DROUGHT ANALYSIS USING CORDEX SIMULATIONS ...etd.lib.metu.edu.tr/upload/12622080/index.pdfTHE DEGREE OF MASTER OF SCIENCE IN CIVIL ENGINEERING APRIL 2018 iii Approval of the thesis:

DROUGHT ANALYSIS USING CORDEX SIMULATIONS OVER THE

MEDITERRANEAN CLIMATE REGIONS OF TURKEY

A THESIS SUBMITTED TO

THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES

OF

MIDDLE EAST TECHNICAL UNIVERSITY

BY

ANIL YILDIRIM POYRAZ

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR

THE DEGREE OF MASTER OF SCIENCE

IN

CIVIL ENGINEERING

APRIL 2018

iii

Approval of the thesis:



submitted by ANIL YILDIRIM POYRAZ in partial fulfillment of the requirements

for the degree of Master of Sciences in Civil Engineering Department, Middle

East Technical University by,

Prof. Dr. Halil Kalıpçılar

Dean, Graduate School of Natural and Applied Sciences

Prof. Dr. İsmail Özgür Yaman

Head of Department, Civil Engineering

Prof. Dr. İsmail Yücel

Supervisor, Civil Engineering Dept., METU

Examining Committee Members:

Prof. Dr. Elçin Kentel

Civil Engineering Dept., METU

Prof. Dr. İsmail Yücel


Assoc. Prof. Dr. M. Tuğrul Yılmaz


Assoc. Prof. Dr. Önder Koçyiğit

Civil Engineering Dept., Gazi University

Assoc. Prof. Dr. Ceylan Talu Yozgatlıgil

Statistics Dept., METU

Date: 27/04/2018

iv

I hereby declare that all information in this document has been obtained and

presented in accordance with academic rules and ethical conduct. I also declare

that, as required by these rules and conduct, I have fully cited and referenced

all material and results that are not original to this work.

Name, Last Name : Anıl Yıldırım POYRAZ

Signature :

v

ABSTRACT



Poyraz, Anıl Yıldırım

M.Sc., Department of Civil Engineering

Supervisor : Prof. Dr. İsmail Yücel

April 2018, 154 pages

Drought has been a significant result of climate change that causes variance on

precipitation regimes. Mediterranean region is one of the hotspots of the world in this

respect. Dry summers and rainy winters -the characteristic of this climate type- makes

this region more vulnerable to the effects of climate change. Hence, it is important to

monitor drought considering the increasing population and economic facilities in the

regions that are under Mediterranean climate conditions in Turkey. This study aims

to assess the trends in drought by applying the Standardized Precipitation Index(SPI)

for 5 timescales – from 1 month to 12 months. The model grid data that corresponds

to meteorological stations distributed from south to west within the study area was

obtained from 12 different Global Circulation Model / Regional Climate Model

couplings of CORDEX project. Observed and modeled prediction data were

compared for reference period (1971-2005) in order to detect the most reliable

models. Afterwards, modified Mann-Kendall trend test was applied on the SPI and

annual precipitation values for the entire period (1972-2100). The trends were

vi

estimated by linear regression for the locations in which Mann-Kendall results

indicated a significant change. In conclusion, a persistent increasing drought trend

was detected for Muğla and western Antalya parts such that all models are coherent.

On the other hand, the divergence of the trends for some regions according to

different models signifies the discrepancy of models. Besides, the drought trends are

decreasing for some regions (especially Southern Marmara) as the timescale

increases.

Keywords: Climate change, drought, Standardized Precipitation Index,

Mediterranean climate region, CORDEX project

vii

ÖZ

CORDEX SİMÜLASYON VERİLERİ KULLANILARAK TÜRKİYE’NİN

AKDENİZ İKLİM BÖLGELERİNDE KURAKLIK ANALİZİ

Poyraz, Anıl Yıldırım

Yüksek Lisans, İnşaat Mühendisliği Bölümü

Tez Yöneticisi : Prof. Dr. İsmail Yücel

Nisan 2018, 154 sayfa

Yağış rejimlerinde değişimlere sebep olan iklim değişikliğinin önemli sonuçlarından

biri de kuraklıktır. Akdeniz Bölgesi bu açıdan Dünyadaki hassas noktalardan biridir.

Bu iklim tipinde yazların kurak, kışların yağışlı olması bölgeyi iklim değikliğinin

etkilerine daha açık hâle getirmiştir. Bu sebeple, Türkiye’nin Akdeniz iklimi etkisi

altındaki bölgelerinde artan nüfus ve ekonomik etkinlikler de düşünüldüğünde

kuraklığı incelemek oldukça önemlidir. Bu çalışmada, Standartlaştırılmış Yağış

İndeksi(SPI) 1 aydan 12 aya kadar 5 zaman ölçeğinde uygulanarak kuraklıktaki

trendler değerlendirilmiştir. 12 farklı iklim modelinin, incelenen alanının güneyinden

batısına yayılmış meteorolojik gözlem istasyonlarının konumuna denk gelen yağış

verileri CORDEX projesinden alınmıştır. En güvenilir modelleri belirlemek amacıyla

referans dönemi (1971-2005) için gözlem ve model yağış verileri karşılaştırılmıştır.

Ardından, düzenlenmiş bir Mann-Kendall testi tüm dönem için (1972-2100) SPI ve

yıllık yağış değerlerine uygulanmıştır. Mann-Kendall testinin anlamlı değişim işaret

ettiği noktalar için lineer regresyon yöntemiyle trendler hesaplanmıştır. Sonuçta,

viii

Muğla ve Batı Antalya için kuraklıkta tüm modellerin sonuçlarının uyum içinde

olduğu ciddi bir artış trendi tespit edilmiştir. Öte yandan, bazı bölgeler için model

sonuçların çeşitliliği, modellerin farklılıklarına işaret etmektedir. Ayrıca, Güney

Marmara başta olmak üzere bazı bölgelerde zaman ölçeği arttıkça kuraklık trendinde

azalış sözkonusudur.

Anahtar kelimeler: İklim değişikliği, kuraklık, Standartlaştırılmış yağış indeksi,

Akdeniz iklim bölgesi, CORDEX projesi

ix

ACKNOWLEDGEMENTS

I would like to express my deepest gratitude to my supervisor Prof.Dr. İsmail Yücel

for his patient guidance and precious advices throughout the research.

I would like to thank my dear labmate Rizwan Aziz for his valuable help which made

this thesis possible.

I thank my friends Ezgi Kıyıcı, Özgün Emre Koç, Çiğdem Aygün and Yasinalp Örsel

for their support throughout thesis study. Lastly, I would like to thank my cat Kumpir

who never left me alone along the study nights.

x

TABLE OF CONTENTS

ABSTRACT ........................................................................................................... v

ÖZ ....................................................................................................................... vii

ACKNOWLEDGEMENTS ...................................................................................ix

TABLE OF CONTENTS ........................................................................................ x

LIST OF FIGURES ............................................................................................. xii

LIST OF TABLES ............................................................................................... xiv

ABBREVIATIONS ............................................................................................. xvi

CHAPTERS

1. INTRODUCTION .............................................................................................. 1

Introduction .................................................................................................. 1

The Significance of the Study ....................................................................... 5

Literature Review ......................................................................................... 6

Description of Thesis .................................................................................. 10

2. DATA, STUDY AREA AND METHODS ........................................................ 11

Data and Study Area ................................................................................... 11

Study Area ........................................................................................... 11

Observation Data .................................................................................. 13

Model Data ........................................................................................... 13

2.2. Methods...................................................................................................... 16

Standardized Precipitation Index .......................................................... 18

Modified Mann-Kendall Test ................................................................ 22

xi

Linear Regression ................................................................................ 23

Statistics Used for Measuring Model Performance ............................... 24

3. RESULTS ........................................................................................................ 25

Performance of the Models ........................................................................ 25

Daily Evaluations................................................................................. 25

Monthly Evaluations ............................................................................ 36

3.2. Effects of Time Duration in SPI Analysis ................................................... 43

3.3. Modeled trend analysis of SPI-3, -12, and annual precipitation at selected

stations ............................................................................................................. 46

3.4. The Slope of Trend Analysis of SPI and Annual Precipitation From Ensemble

Model Results ................................................................................................... 52

SPI 1 .................................................................................................... 53

SPI 3 .................................................................................................... 63

SPI 6 .................................................................................................... 72

SPI 9 .................................................................................................... 79

SPI 12 .................................................................................................. 85

Annual Precipitation ............................................................................ 92

Changes in the Frequency of Drought Between the Mid and End of the Century

............................................................................................................ 101

Areal Extension of Drought within the period .......................................... 115

4. SUMMARY, CONCLUSION AND RECOMMENDATIONS ....................... 119

REFERENCES .................................................................................................. 122

APPENDICES

A. MATLAB CODES .................................................................................. 126

B. TREND SLOPE TABLES ....................................................................... 131

xii

LIST OF FIGURES

FIGURES

Figure 1. The distribution of annual precipitation for Turkey based on the period

1981-2010 (Downloaded from WEB4).................................................................... 2

Figure 2 Classification of drought depending on duration and effects (Downloaded

from WEB7). .......................................................................................................... 4

Figure 3. Illustration of Hadley Circulation (Heffernan, 2016). Dry zones are

extending as Hadley cell is shifting polewards. ....................................................... 7

Figure 4. Mediterranean climate regions on the Earth (Downloaded from WEB8). 12

Figure 5. Climate map of Turkey (Downloaded from WEB9). .............................. 12

Figure 6. Inner square surrounds EURO-CORDEX domain area (Stagge et al., 2015)

............................................................................................................................. 13

Figure 7. The locations of model data for projection period ................................... 16

Figure 8. The steps of the study based on the data and operations used. ................. 17

Figure 9 Example of an equiprobability transformation from a fitted gamma

distribution to the standard normal distribution (Hughes and Saunders, 2002) ....... 20

Figure 10 Obtaining linear regression line ............................................................. 23

Figure 11. Daily mean precipitation of entire area for reference period.................. 27

Figure 12. Seasonally grouped boxplot of monthly precipitation for entire area ..... 37

Figure 13. Plot of monthly precipitation means of observation and models. The bars

on model mean line show the standard deviation gap of 12 models. ...................... 38

Figure 14. Monthly precipitation means of observation and models ...................... 39

Figure 15. SPI values for Muğla station according to model 1-1 predictions (1972-

2100). The linear regression equation is on top-right hand corner. ......................... 44

Figure 16. Annual precipitation for all models with stacked lines .......................... 49

Figure 17 The study area with province names ...................................................... 52

Figure 18 The geographical distribution of trends for SPI 1. The legend is valid for

rest of the SPI trend maps ..................................................................................... 56

xiii

Figure 19. The geographical distribution of trends for SPI 3 ..................................65



Figure 22. The geographical distribution of trends for SPI 12 ................................86

Figure 23. The geographical distribution of trends for annual rainfall ....................94

Figure 24. Changes in the frequency of drought. ‘Mod’, ‘sev’ and ‘ext’ indicate

moderately, severely and extremely dry classes respectively. .............................. 103

Figure 25. The percentage of the stations under dry conditions ............................ 115

Figure 26. The average percentage of the stations under dry conditions for all models

............................................................................................................................ 117

xiv

LIST OF TABLES

TABLES

Table 1. Location of the observation stations (* signifies the locations which were

only studied in projection)..................................................................................... 14

Table 2. Model data list. ‘Model no’ numbering was given to provide a convenience

throughout the thesis. ............................................................................................ 15

Table 3 SPI value interpretation (Mckee et al., 1993) ............................................ 21

Table 4. Correlation coefficient(CORR) and Root mean square error(RMSE) values

for moving averages of daily means ...................................................................... 31


for successive monthly .......................................................................................... 32

Table 6. Correlation coefficient(CORR) and root mean square errors(RMSE) analysis

for seasonally grouped months .............................................................................. 34

Table 7. Trends in annual precipitation for reference period (1971-2005). ↘ denotes

negative slope, ↔ denotes no significant trend, ↗ denotes positive slope. ............. 35

Table 8. The statistics of coupled models for projection period ............................. 36

Table 9. The statistics for coupled models for projection period ............................ 36

Table 10. Trend change table for 5 stations. ↘ denotes negative slope, ↔ denotes no

significant trend, ↗ denotes positive slope. ........................................................... 48

Table 11. Trend slope values for model 1-1 ......................................................... 131

Table 12. Trend slope values for model 1-2 ......................................................... 133

Table 13. Trend slope values for model 1-3 ........................................................ 135






xv

Table 19. Trend slope values for model 3-2 ....................................................... 147




xvi

ABBREVIATIONS

CLMcom Climate Limited-area Modelling Community

CNRM Météo-France/Centre National de Recherches Météorologiques

CORDEX Coordinated Regional Climate Downscaling Experiment

CORR Correlation Coefficient

DMI Danish Meteorological Institute

GCM General Circulation Model/Global Climate Model

IPCC Intergovernmental Panel on Climate Change

IPSL Institut Pierre-Simon Laplace

KNMI Royal Netherlands Meteorological Institute

MGM Turkish State Meteorological Service

MOHC Met Office Hadley Centre

NASA The National Aeronautics and Space Administration

RCM Regional Climate Model

RCP Representative Concentration Pathways

RMSE Root Mean Square Error

SMHI Swedish Meteorological and Hydrological Institute

SPI Standardized Precipitation Index

http://www.cordex.org/

http://www.ipcc.ch/

https://mgm.gov.tr/eng/forecast-cities.aspx?m=ISTANBUL

https://en.wikipedia.org/wiki/Representative_Concentration_Pathways

1

CHAPTER 1

INTRODUCTION

Introduction

Water moves in a cycle on the earth and in the atmosphere. It evaporates from the

surface of the earth, cools and condenses as it rises into atmosphere, and falls again

to the surface in different forms of precipitation (WEB1).

Climate is the typical and averaged weather of a region or a city over many years.

Climate characteristics consist of many attributions from seasonal temperature

differences to precipitation regimes. Changes in climate take several years - decades,

centuries and even millennials (WEB2).

Frequent and precise measurements of any form of precipitation is indispensable to

determine changes in and make models of Earth’s water cycle. In addition to

observations, climate models reproduce observed features of recent climate and past

climate changes by benefiting well-established physical principles to (Randall et al.,

2007). A General Circulation Model (also known as Global Climate Model and

abbreviated as GCM) can provide reliable prediction information on big scales

(around 1000 by 1000km) while Regional Climate Models applied over a limited area

and driven by GCMs can provide information on much smaller scales (WEB3). This

improves assessing the changes in precipitation regimes in many vulnerable regions

of the world.

2

Turkey is a country that has a climatic diversity though it is situated in large

Mediterranean location. This diversity of climatic conditions is mainly due to diverse

nature of the landscape. The mountains in the south and north coast run parallel to

the seashore and therefore marine climate cannot penetrate to the interior parts. Only

the western parts are relatively more open to marine effects since the mountains are

not parallel to the shore, but perpendicular.

The climatic differences between regions of Turkey can be figured out by

geographical distribution of annual precipitation (

Figure 1) Annual rainfall along Mediterranean Sea and Aegean Sea varies from 580

to 1,300 millimeters, depending on location whilst Black Sea coast receives the

highest annual rainfall (Sensoy et al., 2016). The amount of rainfall decreases

gradually to the inland. Only a small part from the east of the country receives

precipitation as much as coastline.

Figure 1. The distribution of annual precipitation for Turkey based on the period

1981-2010 (Downloaded from WEB4)

3

The increasing concentration of greenhouse gases leads the atmosphere to be warmer

as a result of trapped solar energy. This phenomenon is called ‘global warming’

which refers to global temperature rises for long-term, whereas ‘climate change’ is a

broader term that indicates not only the changes in averages but also increases in

occurrence of extreme events like floods, drought and heatwaves or changes in rain

and snow patterns (WEB2).

Drought is a bit latent phenomenon comparing to other extreme weather events since

it is not as instantaneous as floods that emerge in minutes or heatwaves that evokes

itself immediately. However, it is such an event that has widespread and long-termed

effects to nature and society. Even, mass migration of people is one of the striking

results of these effects (Raleigh et al., 2008). The cost of drought must also be taken

into account at evaluation of its damage. The estimation of the cost of a four-year

drought that hit California is 2.7 billion US Dollars in 2015, according to a study

from University of California-Davis (WEB5). At the beginning of 2018, an increase

in electricity price was discussed due to decreases in electricity production of

hydropower plants in consequence of drought in Turkey (WEB6).

Since precipitation is a vital component of water cycle, the changes in climate directly

affect the spatial and temporal distribution and quantity of precipitation. Assessing

the changes and trends in precipitation and concluding about drought is an arguable

issue as well as essential. This complexity arises from the problem of defining

drought. Redmond(2002) explains drought simply ‘insufficient water to meet needs’

following a discussion on the approaches to the phenomenon. This definition

highlights the importance of both the supply and the demand sides of the subject

(Redmond, 2002).

Wilhite and Glantz (1985) endeavored to categorize the drought. They categorized

the definitions into four: meteorological, hydrological, agricultural,

and socioeconomic.

4

Meteorological definitions are the most common and define drought usually based

on the degree of dryness and the duration of the dry period. Definitions of

meteorological drought must be considered separately for every region since the

meteorological conditions that result in deficiencies of precipitation are highly

irregular from region to region.

Agricultural drought associates numerous characteristics (precipitation shortages,

soil water deficits etc.) of meteorological drought to agricultural impacts. A useful

definition of agricultural drought should consider the variable susceptibility of crops

at different stages of crop growth.

Figure 2 Classification of drought depending on duration and effects (Downloaded

from WEB7).

5

Hydrological drought is linked with the effects of duration of precipitation shortfalls

on surface or subsurface water supply (WEB7). This type of droughts is mostly out

of phase with both meteorological and hydrological droughts since it takes longer to

show up precipitation deficiencies on hydrological system components such as soil

moisture, streamflow, and groundwater and reservoir levels.

A variety of indices have been proposed and used to assess drought up to now.

However, developing an index to assess drought is an inseparable matter from

defining drought. Thus, quantifying drought by indices is a difficult geophysical

endeavor. Standardized Precipitation Index(SPI) is a meteorological drought index

that is widely used to detect drought for different timescales. Keyantash and Dracup

(2002) finds out that SPI is the most successful index in measuring drought right after

rainfall deciles. They analyze 14 types of indices from all drought forms

(meteorological, hydrological and agricultural) based on six criteria: robustness,

tractability, transparency, sophistication, extendability, and dimensionality. SPI is

also distinguished with its ability to measure the severity of drought and selected to

measure drought in this study owing to all these features of it.

The Significance of the Study

The main goal of this study is to investigate the drought conditions from past to the

end of 21st century in Mediterranean climate region of Turkey that is most vulnerable

to the effects of climate change because of the increase in temperature and decrease

in precipitation (Dabanlı et al., 2017; Topcu et al., 2010). The drought analyses were

performed by calculating the well-known SPI values for drought at 1, 3, 6, and 12

months timescales, assessing the impact of drought at different levels, i.e.

meteorological, hydrological and agricultural droughts. Ensemble modeling

approach releases 12 GCM/RCM pairs from CORDEX (the Coordinated Regional

Climate Downscaling Experiment) project was used to make drought predictions not

only for the past but also future period till the end of century. First, in reaching the

6

main goal of this study, the performance analyses of the GCM/RCMs pairs in

estimating monthly precipitation that are used to derive SPI indices were made at 46

grid locations corresponding to meteorological stations distributed to the study area.

Second, the Mann Kendall trend test was applied to SPI values calculated through the

period from 1972 to 2100 for each model pair. Finally, the assessment of drought at

various magnitudes was performed at locations where drought is statistically

significant from multi-model system over entire study area. As a result, the

consistency of drought that appears within the region by the end of century was

documented with the support of ensemble model approach.

Literature Review

IPCC reports (2013) revealed that decreases in soil moisture and increases in

agricultural drought are likely in presently dry regions by the end of 21st century

according to the projections from regional to global scale under RCP8.5 scenario.

The drying in soil moisture is also consistent with projected changes in Hadley

circulation (Figure 3) and surface temperature increases in Mediterranean,

Southwestern US and Southern African regions (IPCC, 2013). In addition to this,

Giorgi (2006) highlighted the vulnerability of Mediterranean and North Eastern

European regions by defining them the climate change hot-spots.

An extensive research that was conducted by Cook et al. (2016) investigated the

drought for the whole region around Mediterranean Sea. They analyzed the drought

variability for 900 years (1100-2012) in the Old World Drought Atlas (OWDA), a

spatiotemporal tree ring reconstruction of the June-July-August self-calibrating

Palmer Drought Severity Index. The outcomes indicated an east-west coherence in

drought on multidecadal and centennial timescales. However, the analysis results of

Levant region (Cyprus, Israel, Jordan, Lebanon, Palestine, Syria, and Turkey)

indicated that recent dry extremes are extraordinary during the last millennium. This

7

provided a support to studies claiming that the anthropogenic climate change has a

significant effect (Cook et al., 2016).

Figure 3. Illustration of Hadley Circulation (Heffernan, 2016). Dry zones are

extending as Hadley cell is shifting polewards.

Many studies have been done based on the evaluation of the performance of the

climate models and making predictions depending upon them. The followings

provide a summary of the results of some of these studies.

Giorgi and Lionello (2008) determined that GCM and RCM simulations are generally

similar to each other at large scale. Though, it was pointed that precipitation change

signal produced by RCMs also take into the orographically effects account.

8

Leng et al. (2015) investigated climate change impacts on drought in China. They

detected the different response of same models (HadGEM2-ES and MIROC-ESM-

CHEM) for different drought types. Predicting more extreme droughts than mean

droughts is another significant result of their study.

Osuch et al. (2016) analyzed SPI and rainfall trends for Poland with RCM projections

from ENSEMBLES project. These analyses were conducted for a period that consists

both historical and future data: 1971-2099. Bias correction was applied to model data,

though, the trends in SPI slightly changed after correction. Considering the

consistency of the models, it was detected that the results are similar for some models

for the study area. Modified Mann-Kendall test was used for trend detection in this

study. Lee et al. (2017) determined the influence of climate change and possible rises

on drought conditions for Hwanghae Plain in Korea with regionally downscaled data.

Stagge et al. (2015) investigated future meteorological drought based on CORDEX

data for RCP 4.5 and RCP 8.5 scenarios. They highlighted the conflicting results of

previous studies on drought severity even though their consistency on regional

hotspots and CORDEX project’s capacity to improve the reliability and consistency

of the analyses since the projections have been processed at a much finer resolution.

The results of this study indicated significant increases in meteorological drought

frequency and severity for Mediterranean region while areal extensions were likely

for Atlantic coast and Southern Europe according to period 1971-2000. It was also

detected that the changes in the occurrence of moderate and severe droughts and the

affected area were regardless of the long-term emission scenarios for the near term

(2011-2040). Additionally, the increase scenarios in drought are consistent around

Mediterranean for both scenarios.

Kara et al. (2016) investigated climate change impacts on extreme precipitation of

Ömerli catchment in İstanbul by using ensemble climate modeling. The ensembles

of daily precipitation time series from 15 different RCMs driven by 5 different GCMs

under A1B climate change projections obtained through EU-ENSEMBLES project

for two periods: reference (1960-1990) and future (2071-2100). An increase in

9

extreme precipitation in winter, spring and summer is expected while a decrease in

autumn is likely according to the results. They also used the geographically weighted

regression (GWR) method to downscale climate change impacts to this small

catchment and GWR provided significant modifications to these changes and agreed

on the direction of change from RCMs.

There are many studies dedicated to analyze the drought for entire Turkey and certain

basins of Turkey’s Mediterranean climate region as well. Sönmez et al. (2005) found

out that drought vulnerability of Turkey for varying time steps portrayed diverse but

consistent picture. Their study also revealed varying trends for different regions in

terms of drought severity and duration. While the southeastern and eastern parts of

Turkey are more open to moderate droughts at short timescales, the impact would be

anticipated less at the coastal part since the droughts are only effective at longer

durations and occur at moderate levels. Nevertheless, coastal and interior parts more

tend to occur severe droughts. These facts bring negative consequences for different

sectors that needs water for varying periods of the year. Interior parts will suffer from

agricultural drought whilst hydrological drought will occur at longer time steps at the

coastal parts.

Türkeş (2012) revealed briefly the effects of climate change in a study that examines

the observed and projected drought and desertification in Turkey. The effects of

global warming were considered with evaluating the changes in extremes in this

study. It is important that the results based on a modified standardized precipitation

index(MSPI) showed increase in drought severity for the regions under

Mediterranean climatic conditions and the inland parts of the country which is

neighbor to this climate region. The vulnerability of Turkey with respect to intensive

and broad winter droughts -which are related to high positive modes of North Atlantic

Oscillation- is critical as well.

Unlike the other studies that are dedicated to determine the effects of climate change,

Türkeș et al. (2016) also inquired if the climate of Turkey is really changing in a study

10

that compared two consecutive time periods: 1950-1980 and 1981-2010. They

detected some variations in the present geographical patterns of climate regions.

Increasing precipitation amounts in the northern and eastern regions in contrast to

decreasing amounts in the west, central and southern regions were serious outcomes

of this study.

Gümüş and Algın (2017) examined the relation between meteorological and

hydrological drought for Seyhan-Ceyhan River Basins using SPI and SDI

(Streamflow Drought Index). They found that a meteorological drought demonstrates

hydrological drought for the following year. This result is crucial for water

management with more frequent and severe droughts.

Description of Thesis

In the first chapter of the thesis, a brief information about climate change, GCM-

RCM simulations and drought indices are given. The previous studies are also

mentioned in this chapter. Details about data, study area and methods are explained

in the following chapter. The third chapter presents the results of analysis. The results

are discussed in chapter 4. The last chapter provides the summary, conclusions and

recommendations.

11

CHAPTER 2

DATA, STUDY AREA AND METHODS

Data and Study Area

Study Area

Mediterranean climate is a major climate type of the Köppen classification and

characterized by hot, dry summers and cool, wet winters. The regions which under

this climatic condition are located between about 30° and 45° latitude north and south

of the Equator and on the western sides of the continents (Kottek et al., 2016). Figure

4 shows five Mediterranean climate regions on the Earth.

This region presents several aspects of interest, such as its important inter-annual

variability in precipitation and temperature, and the severe economic

damages and losses of life due to droughts, flooding events or heat or cold waves

occurred in the last decades, together with an increase in population and infrastructure

(Easterling et al., 2000). In this study, the climatic conditions were mainly taken into

consideration at determination of study area rather than other identifiers like regional

or provincial borders. Climate frontiers are not very certain and differ a bit from one

map to other for Turkey. Still, the maps are consistent in general. Climate regions

showed in

Figure 5 were taken as a basis at determination of study area in the thesis. Only a

small part consisting of the north and east coast of the Marmara Sea was not studied.

12

Figure 4. Mediterranean climate regions on the Earth (Downloaded from WEB8).

Figure 5. Climate map of Turkey (Downloaded from WEB9).

13

Observation Data

Observed precipitation data for monthly rainfall was obtained from MGM (Turkish

State Meteorological Service) for 46 stations to evaluate the performance of model

data (Table 1). This evaluation period was also called ‘reference period’, which refers

to the period 1971-2005. The stations were selected homogeneously for the whole

study area as much as possible (Figure 7).

Model Data

Model data was selected from Eur11 domain (Figure 6), that includes data in

0.11degree (~12.5 km) resolution. This is the highest resolution produced in

CORDEX project (WEB10).

Figure 6. EURO-CORDEX domain area surrounded by the inner square (Stagge et

al., 2015)

14

Table 1. Location of the observation stations (* signifies the locations which were

only studied in projection)

Station Name Latitude Longitude Station Name Latitude Longitude

BANDIRMA 40.35 27.97 KARAİSALI 37.27 35.07

AYVALIK 39.32 26.70 MANAVGAT 36.78 31.43

DİKİLİ 39.07 26.88 ERDEMLİ 36.62 34.30

AKHİSAR 38.92 27.85 CEYHAN 37.03 35.82

KUŞADASI 37.87 27.25 DÖRTYOL 36.85 36.22

DİDİM* 37.48 27.27 ISLAHİYE 37.03 36.63

BODRUM 37.05 27.43 GAZİPAŞA 36.27 32.32

DALAMAN 36.75 28.78 YUMURTALIK 36.77 35.78

ANAMUR 36.08 32.83 SAMANDAĞ 36.08 35.97

SİLİFKE 36.38 33.93 ACIPAYAM 37.42 29.33

İSKENDERUN 36.58 36.17 TEFENNİ 37.32 29.77

FİNİKE 36.30 30.15 GEMLİK* 40.44 29.15

KAŞ* 36.20 29.65 KARACABEY* 40.13 28.33

SALİHLİ 38.48 28.13 MUDANYA* 40.37 28.90

SEFERİHİSAR 38.35 26.83 M.KEMALPAŞA* 40.04 28.40

ÖDEMİŞ 38.23 27.97 AYVACIK* 39.61 26.40

NAZİLLİ 37.92 28.32 OSMANIYE* 37.08 36.25

ELMALI 36.75 29.92 ALANYA* 36.55 32.025

MUT 36.65 33.43 MANISA 38.62 27.43

KARATAŞ 36.57 35.38 IZMIR 38.43 27.17

*MENEMEN 38.58 27.07 AYDIN 37.85 27.85

FETHİYE 36.62 29.12 DENIZLI 37.78 29.08

MARMARİS 36.85 28.27 MUGLA 37.22 28.37

BURHANİYE* 39.50 26.98 ANTALYA 36.88 30.7

MİLAS 37.32 27.78 MERSIN 36.8 34.6

YATAĞAN 37.35 28.13 ADANA 37 35.33

KOZAN 37.45 35.82 ANTAKYA 36.2 36.17

DATÇA 36.75 27.67 BALIKESİR* 39.63 27.88

KÖYCEĞİZ* 36.97 28.68 CANAKKALE 40.15 26.42

KORKUTELİ 36.75 30.20 BURSA 40.18 29.07

15

Model data was obtained for 12 models from CORDEX project (Table 2). RCM-

GCM couplings was made of 4 different Global Climate Models (in other words

Driving Models) and 6 Regional Climate Models. The producers of the RCMs were

stated for information purposes.

Table 2. Model data list. ‘Model no’ numbering was given to provide a convenience

throughout the thesis.

Model No GCM Institute RCM

1-1

ICHEC-EC-EARTH

DMI HIRHAM5

1-2 CLMcom CCLM4-8-17

1-3 KNMI RACMO22E

1-4 SMHI RCA4

2-1

CNRM-CERFACS-CNRM-CM5

CNRM ALADIN53

2-2 CLMcom CCLM4-8-17

2-3 SMHI RCA4

3-1

MOHC-HadGEM2-ES

CLMcom CCLM4-8-17

3-2 KNMI RACMO22E

3-3 SMHI RCA4

4-1 IPSL-IPSL-CM5A-MR

SMHI RCA4

4-2 IPSL-INERIS WRF331F

The model data was extracted for 60 locations (Figure 7) from CORDEX data grids.

46 of these were used in performance analysis since reliable data could be obtained

for 46 observation stations. On the other hand, all of 60 locations were analyzed for

future projection. RCP 8.5 - the scenario of highest greenhouse gas emission- was

considered for future in the study.

16

Figure 7. The locations of model data for projection period

2.2. Methods

In this part, the process of analysis was introduced. Estimation methods that were

applied to precipitation data and SPI values are stated as well. Figure 8 describes the

analysis steps.

17

Figure 8. The steps of the study based on the data and operations used.

18

Standardized Precipitation Index

The Standardized Precipitation Index is a meteorological drought index that was

introduced by Mckee et al. (1993). It interprets observed precipitation as a

standardized departure with respect to a rainfall probability distribution function

(Keyantash and Dracup, 2002). The calculation of SPI value for desired period is

based on the long-term precipitation record. Guttman (1999) recommends at least 50

years of data for a reliable calculation.

The calculation of SPI begins with modeling the monthly precipitation time series

using different statistical distributions (Lloyd-Hughes and Saunders, 2002). The first

is the gamma distribution, whose probability distribution is defined as

g(x) =1

βαΓ(α)xα−1e−x/β for x > 0 (1)

where, α > 0 is shape parameter, β > 0 is scale parameter, and x > 0 is the amount of

monthly precipitation. Γ(α) is the gamma function, which is defined as

Γ(α) = limn→∞

∏n! ny−1

y + υ≡ ∫ ya−1

∞

0

n−1

υ=0

e−ydy

(2)

Fitting the distribution to the monthly precipitation data requires estimating α and β.

Edwards and McKee (1997) suggest using the approximation of Thom (1958) to

estimate these parameters as follows:

α =1

4A(1 + √1 +

4A

3)

(3)

β =

x

α

(4)

19

where, for n observations

Α = ln(x) −

∑ ln (x)

n

(5)

The expression of cumulative probability G(x) of an amount of precipitation

occurring for a given month and timescale is yielded by integrating the probability

density function with respect to x and inserting the estimates of α and β:

G(x) = ∫ g(x)dx =

1

βαΓ(α)

x

0

∫ xαx

0

e−x β⁄ dx

(6)

Substituting t for x/^β reduces Equation (6) to

G(x) =1

Γ(α)∫ tα−1e−1dt

x

0

(7)

which is the incomplete gamma distribution function. Since the gamma distribution

is undefined for x = 0, and q = P(x = 0) > 0 where P(x = 0) is the probability of zero

precipitation, the cumulative distribution becomes

H(x) = q + (1 − q)G(x) (8)

The cumulative probability distribution is then transformed into the standard normal

distribution to obtain the SPI value. This process is illustrated in Figure 9. The first

panel shows the empirical cumulative probability distribution for a 3-month average

December–January–February (DJF) of precipitation over the south east of England

for the period 1901–99. Over-plotted is the theoretical cumulative probability

distribution of the fitted gamma distribution. The second panel indicates the standard

normal cumulative probability. To convert a given precipitation level to its

corresponding SPI value, first locate the precipitation value on the abscissa of the

left-hand panel, draw a perpendicular, and locate the point of intersection with the

20

theoretical distribution. Then project this point horizontally (maintaining equal

cumulative probability) until it intersects with the graph of standard normal

cumulative probability. The intersection between a line drawn vertically downward

from this point and the abscissa determines the SPI value (1.1 in this example for 77

mm precipitation).

The above approach is not practical for calculating the SPI for large numbers of data

points. The approximate conversion provided by Abramowitz and Stegun (1965) can

be employed as an alternative following Edwards and McKee (1997):

Z = SPI = − (t −

c0 + c1t + c2t2

1 + d1t + d2t2 + d3t3) for 0 < H(x) ≤ 0.5

(9)

Z = SPI = + (t −

c0 + c1t + c2t2

1 + d1t + d2t2 + d3t3) for 0.5 < H(x) ≤ 1

(10)

Figure 9 Example of an equiprobability transformation from a fitted gamma

distribution to the standard normal distribution (Hughes and Saunders, 2002)

21

In this thesis, the index values were obtained via a MATLAB code that is stated in

Appendix A.

McKee and others used a classification system based on SPI values to define drought

intensities as shown in Table 3 . It extends from extremely wet to extremely dry.

Table 3 SPI value interpretation (Mckee et al., 1993)

The SPI can be computed for any chosen timescales from 1 month to 48 months. This

flexibility is a powerful feature of the SPI that provides useful information unless we

have a clear idea of the desired intervals. In this study SPI was calculated for five

intervals (1,3,6,9 and 12 months) for given periods. Afterwards, trends of SPI values

were evaluated for these five intervals consecutively.

It should be noticed that SPI values were shown and analyzed since 1972 instead of

1971. This difference arose from the nature of SPI calculation. To obtain the SPI 12

value the former 12 months precipitation data is required. Namely, the first SPI 12

value was calculated for the first month of 1972. Then, the SPI values at shorter

timescales were also assessed from 1972 to provide consistency with SPI 12.

22

Modified Mann-Kendall Test

Mann Kendall test (Mann, 1945 ; Kendall, 1975) is one of the widely used non-

parametric tests for detecting trends in time series. The Mann-Kendall trend test is

derived from a rank correlation test for two groups of observations proposed by

Kendall (1975). The correlation between the rank order of the observed values and

their order in time is the key part of Mann-Kendall trend test. However, a modified

Mann– Kendall test has been developed in order to avoid problems with

autocorrelation (Hamed and Ramachandra Rao, 1998). The Mann-Kendall test

statistics S calculated from the following equation:

S = ∑ ∑ sgn(xj − xk) =

n

j=k+1

n−1

k=1

{

+1 if (xj − xk) > 0

0 if (xj − xk) = 0

−1 if (xj − xk) < 0

(11)

Where, number of data is n. Additionally, the correction ratio n/nS* is introduced

during the calculation of a variance of the S statistics to account for an effect of serial

correlation.

var∗(S) = var(S)n

nS∗

(12)

n

nS∗ = 1 +

2

n(n − 1)(n − 2)∑(n − i)(n − i − 1)(n − i

n−1

i=1

− 2)ρS(i)

(13)

where, ρS denotes the autocorrelation function.

Since we are detecting the trend in SPI values, it is important to use such a modified

trend test that considers serial correlation. This test was also applied to annual

precipitation in order to detect any relation with drought. The significance level was

23

taken 0.05 (95% confidence level) for the all trend tests performed in this study. The

MATLAB code that was used in calculation is stated at Appendix A.

Linear Regression

The slopes of the trendlines were obtained by linear regression. This linear approach

models the relationship between a response variable y (SPI and rainfall) and one or

more explanatory variables (years) denoted x. Since negative values of SPI denotes

dry conditions, negative slope means increasing drought.

Figure 10 illustrates how a regression line is fitted to variables. The predicted values

of y are denoted by ŷ whose equation includes two constants: intercept (w0) and slope

(w1).

y = w0 + w1x (14)

Figure 10 Obtaining linear regression line

24

Statistics Used for Measuring Model Performance

Root mean square error (RMSE) values of the model precipitation data were

estimated for reference period to analyze the reliability of the models. It is one of the

methods that are used to measure the closeness of model data to real values along

with correlation coefficient (CORR). Schaller et al. (2011) evaluate climate models

by ranking their RMSE and CORR values. They use an updated version of the model

ranking performed by Reichler and Kim (2008).

RMSE is calculated by following equation which is based on the difference between

observed and model data values:

RMSE = √∑ (Pi − Oi)

2ni=1

n

(15)

where, O denotes amount of observed precipitation while P denotes predicted

precipitation and n denotes number of data.

The correlation coefficient of two random variables is a measure of their linear

dependence. If each variable has n scalar observations, then the Pearson correlation

coefficient (r) is defined as

r =

∑ (xi − x)(yi − yni=1

√∑ (xi − x)2ni=1 √∑ (yi − y)2n

i=1

(16)

where, xi and yi are the single samples (observation and model values) indexed

with i; x and y are the sample means.

25

CHAPTER 3

RESULTS

Performance of the Models

In the first step of analysis, the observed and model precipitation data were compared

for the reference period 1971-2005. The averages of daily and monthly precipitation

were compared at the same time as long term statistics were calculated and evaluated.

Daily Evaluations

The plots in Figure 11 demonstrate the mean precipitation of each day of year for

reference period. In addition, 31-days moving averages were plotted for both

observation and model data.

Daily averages of model 1-1 well correlated with observation means (Figure 11a).

However, summer and autumn daily means slightly underestimated observation as

well as partially underestimated observed daily mean precipitation for winter and

spring seasons. Model 1-2 diverged from observation averages for almost the entire

year (Figure 11b). This model overestimated observation for spring and summer days

while underestimated in autumn and winter. Model 1-3 correlated with observation

for autumn (Figure 11c). Still, the daily averages of spring overestimated observation.

Summer and winter are the two seasons which daily averages of model 1-3 partially

overestimated and underestimated observation. Summer, autumn and winter daily

means of model 1-4 diverged from observation as well as model correlated with

observation means overall (Figure 11d).

26

Model 2-1 and 2-2 greatly diverged from observed means (Figure 11e and Figure

11f). Both models overestimated summer precipitation. Overestimation carried on for

model 2-2 autumn daily means while model 2-1 underestimated observation winter

daily averages. Model 2-3 well correlated with observation though it is forced by

same GCM with model 2-1 and 2-2 (Figure 11g). Besides, the mean values are very

close to observation at model 2-3.

Model 3-1 correlated with observation to some extent (Figure 11h). Despite, it

overestimated observation means for winter, spring and autumn seasons. Model 3-2

greatly overestimated observation for winter, spring and summer seasons (Figure

11i). Model 3-3 (Figure 11j) well correlated with observation like models 1-4 and 2-

3 which are forced by same RCM: RCA4. Though, model 3-3 underestimated

observation daily means for autumn and winter seasons.

Model 4-1 underestimated observation daily averages for entire year (Figure 11k).

The divergence is greater for winter and spring seasons. Model 4-2 did not correlate

with observation as well as the model daily means greatly overestimated observation

for spring and summer seasons (Figure 11l).

27

Figure 11. Daily mean precipitation of entire area for reference period.

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

a) Model: 1-1

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

b) Model: 1-2

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

c) Model: 1-3

28

Figure 11. (cont’d)

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

d) Model: 1-4

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

e) Model: 2-1

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

f) Model: 2-2

29


0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

g) Model: 2-3

0

1

2

3

4

5

6

7

8

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

h) Model: 3-1

0

1

2

3

4

5

6

7

8

9

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

i) Model: 3-2

30


0

1

2

3

4

5

6

7

1

14

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Da

ily m

ea

n p

recip

. (m

m)

Days of year

j) Model: 3-3

0

1

2

3

4

5

6

7

114

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

k) Model: 4-1

0

1

2

3

4

5

6

7

1

14

27

40

53

66

79

92

105

118

131

144

157

170

183

196

209

222

235

248

261

274

287

300

313

326

339

352

365

Daily m

ean p

recip

. (m

m)

Days of year

l) Model:4-2

31

Table 4 shows CORR and RMSE values for moving averages based on daily means

of the models. Predictions of three models (1-1, 2-3, 1-4) are well correlated with

observed values according to statistics. Daily mean values of model 1-3 predictions

are not far from observed means while correlation is relatively low for this model.

For model 4-1, analysis of daily means revealed opposite situation: The values

differed greatly for all days of the year while followed similar seasonal variation with

observed values (Figure 11k and Table 4).


for moving averages of daily means

Rank Model No CORR Model No RMSE

1 1-1 0.9923 2-3 0.27

2 2-3 0.9908 1-1 0.31

3 1-4 0.9795 1-3 0.45

4 4-1 0.9734 1-4 0.46

5 3-2 0.9565 3-3 0.49

6 3-1 0.9458 1-2 0.56

7 1-3 0.9349 3-1 0.63

8 3-3 0.9264 4-1 0.76

9 1-2 0.8908 2-2 0.79

10 2-2 0.8895 4-2 0.82

11 4-2 0.8663 3-2 0.87

12 2-1 0.5198 2-1 1.08

Table 5 and Table 6 show the CORR and RMSE values for monthly data. 420

monthly precipitation values (from 1971 to 2005) were ordered and analyzed

successively to obtain the outcomes in Table 5. These monthly values were grouped

seasonally and analyzed to obtain Table 6 as well. Since correlation coefficients are

32

relatively low for a reliable evaluation, RMSE values have been discussed rather than

CORR values.


for successive monthly

Rank Model No CORR Model No RMSE

1 3-2 0.3976 4-1 72.58

2 2-3 0.3900 1-4 73.62

3 1-1 0.3881 2-3 74.87

4 3-3 0.3857 1-2 76.62

5 1-4 0.3828 1-3 77.10

6 4-1 0.3768 3-3 78.89

7 3-1 0.3647 1-1 79.02

8 1-3 0.3242 3-1 83.87

9 1-2 0.2946 2-1 84.90

10 2-2 0.2434 2-2 85.00

11 4-2 0.2148 3-2 88.10

12 2-1 0.1434 4-2 90.23

Seeing the RMSE ranking in Table 5, the difference of RMSE values between rank

7 and 8 can be considered as a threshold. The values increased almost with a 1.00

mm/month interval till 7th ranked model while a 4.00 mm/month increase occurred at

8th rank. Models 1-1, 1-4, 2-3, 3-3 and 4-1 are distinguished from these 7 top models

in RMSE ranking along with their high ranks in CORR ranking.

Outcomes of daily means statistics indicated both consistency and divergency for

different models. Models 1-4, 2-3 and 3-3 showed consistency with observation

values according to daily means as well as they are in top ranks in RMSE analysis

33

(Figure 11d, Figure 11g, Figure 11j and Table 4) . However, it should be noticed that

a few models from top ranks in Table 4 diverged from observation daily means

especially for spring season. RMSE values for spring months also indicated this

divergency: models 1-1, 1-2 and 1-3 ranks at 7th, 11th and 8th place for spring

respectively (Table 6). However, the seasons can be ordered as autumn, spring,

summer and winter in terms of the CORR values of models. The models projected

the best performance during autumn and the worst in winter.

34

Table 6. Correlation coefficient(CORR) and root mean square errors(RMSE)

analysis for seasonally grouped months

Rank

Winter Spring

Model

No CORR

Mo

del

No

RMSE Model

No CORR

Mod

el

No

RMSE

1 2-3 0.1056 2-1 102.61 2-3 0.1974 4-1 60.36

2 3-1 0.1046 2-2 102.61 1-1 0.1568 2-3 62.01

3 3-2 0.0806 4-1 102.76 3-3 0.1401 3-3 62.93

4 2-2 0.0795 1-2 103.02 4-1 0.1320 1-4 63.54

5 2-1 0.0699 1-4 104.75 1-4 0.1202 3-2 64.52

6 1-4 0.0585 2-3 105.04 3-1 0.1126 3-1 65.88

7 1-1 0.0546 3-1 109.41 3-2 0.1033 1-1 65.99

8 3-3 0.0448 1-3 109.83 1-3 0.0945 1-3 69.25

9 4-1 0.0431 4-2 113.29 2-2 0.0932 2-1 69.50

10 1-2 0.0091 3-3 113.73 1-2 0.0536 2-2 72.28

11 1-3 -0.0242 1-1 115.61 2-1 0.0388 1-2 72.32

12 4-2 -0.0375 3-2 124.26 4-2 0.0282 4-2 77.04

Rank

Summer Autumn

Model

No CORR

Mo

del

No

RMSE Model

No CORR

Mod

el

No

RMSE

1 2-1 0.1613 1-3 25.09 3-2 0.3136 4-1 69.81

2 3-1 0.1324 1-1 25.13 3-3 0.3055 1-4 70.99

3 1-3 0.1230 1-4 26.16 2-3 0.3021 1-3 74.17

4 3-2 0.1082 4-1 29.31 1-1 0.2859 2-3 74.19

5 1-2 0.1080 3-3 29.40 1-4 0.2642 1-1 74.67

6 4-1 0.0906 1-2 32.04 4-1 0.2629 1-2 74.93

7 2-2 0.0867 3-2 32.16 1-3 0.2516 3-3 78.61

8 1-1 0.0867 2-3 33.11 3-1 0.2294 2-1 79.07

9 4-2 0.0850 3-1 39.38 4-2 0.1904 2-2 88.04

10 2-3 0.0828 2-2 66.33 1-2 0.1896 4-2 88.70

11 1-4 0.0519 4-2 66.85 2-2 0.1895 3-1 96.44

12 3-3 0.0296 2-1 68.96 2-1 0.1448 3-2 97.78

35

Table 7 indicates the trends in annual rainfall for five locations for reference period.

These locations were selected since they are well representing the study area. Models

1-1, 2-2 and 4-2 predicted significant trend for few stations though no significant

trend was detected according to observation and rest of the models. This result can

be considered as a negative outcome on reliability of these three models. Though,

Table 7 indicates that the models are generally trustable in trend analysis.

Table 7. Trends in annual precipitation for reference period (1971-2005). ↘ denotes

negative slope, ↔ denotes no significant trend, ↗ denotes positive slope.

Obs 1-1 1-2 1-3 1-4 2-1 2-2 2-3 3-1 3-2 3-3 4-1 4-2

Adana ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↗

Antalya ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔

Balıkesir ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔

İzmir ↔ ↘ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔

Muğla ↔ ↘ ↔ ↔ ↔ ↔ ↘ ↔ ↔ ↔ ↔ ↔ ↔

Table 8 and Table 9 show the mean and standard deviation values of daily

precipitation simulated by different GCM/RCM pairs for period 1971-2100. These

two tables enable comparing the initiative effects of GCMs and RCMs.

The mean standard deviation of various RCMs with the same GCM model is 1.804

(Table 8) while the mean deviation value of different GCMs with the same RCM is

1.644 (Table 9). This implies the greater impact from RCM models than that from

GCM models on daily precipitation. The uncertainty that comes from RCM models

is higher than that from GCM models in daily precipitation simulation during period

1971-2100.

36

Table 8. The statistics of coupled models for projection period

Model

No GCM RCM mean

mean of

means

standard

deviation

mean

of s.d.

s.d. of

s.d.

1-2 ICHEC-EC-

EARTH

CCLM4-8-17 1.697

1.708

1.491

1.804

0.129572 1-3 RACMO22E 1.913 1.727

1-4 RCA4 1.515 1.516

3-1 MOHC-

HadGEM2-ES

CCLM4-8-17 2.193

2.160

1.932

0.263024 3-2 RACMO22E 2.526 2.329

3-3 RCA4 1.761 1.832

Table 9. The statistics for coupled models for projection period

Model

No GCM RCM mean

mean of

means

standard

deviation

mean

of s.d.

s.d. of

s.d.

1-2 ICHEC-EC-EARTH

CCLM4-

8-17

1.697

2.079

1.491

1.644

0.240143 2-2 CNRM-CERFACS-

CNRM-CM5 2.348 1.546

3-1 MOHC-HadGEM2-ES 2.193 1.932

1-4 ICHEC-EC-EARTH

RCA4

1.515

1.691

1.516

0.174796 2-3

CNRM-CERFACS-

CNRM-CM5 1.796 1.544

3-3 MOHC-HadGEM2-ES 1.761 1.832

Monthly Evaluations

Figure 12 shows boxplots of monthly precipitation from all models and observation

for each season. Each box includes values from 46 locations for reference period. In

addition to this, the mean of the models for monthly precipitation values including

whole stations were shown in Figure 13. The bars around model means denote the

standard deviations of 12 model pairs.

37

Fig

ure

12.

Sea

sonall

y g

rouped

bo

xp

lot

of

mo

nth

ly p

recip

itat

ion f

or

enti

re a

rea

38

The ensemble mean of monthly precipitation values is very near to observation for

most of the months (Figure 13). However, monthly rainfall predictions of a few

models caused a significant divergence of model means from observation means for

the months January, February, May, June and December.

Figure 13. Plot of monthly precipitation means of observation and models. The bars

on model mean line show the standard deviation gap of 12 models.

It can be distinguished which models caused such a divergence at model means by

Figure 14. Models 2-1, 2-2 and 4-2 overestimated summer rainfalls while 2-1

underestimated winter rainfalls. 3-2 overestimated autumn and winter rainfalls as

well. Models 2-1, 2-2, and 4-2 yielded highly meaningless monthly precipitation from

winter months to summer months when they were compared with observed

precipitation.

39

Figure 14. Monthly precipitation means of observation (Obs) and models (Model)

0

20

40

60

80

100

120

140

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month

ly P

recip

itation (m

m)

a) Model 1-1

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

b) Model 1-2

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

c) Model 1-3

40


0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

d) Model 1-4

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

e) Model 2-1

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

f) Model 2-2

41


0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

g) Model 2-3

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

h) Model 3-1

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

i) Model 3-2

42


0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

j) Model 3-3

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

k) Model 4-1

0

20

40

60

80

100

120

140


Month

ly P

recip

itation (m

m)

l) Model 4-2

43

3.2. Effects of Time Duration in SPI Analysis

The plots in Figure 15 show the successive SPI values for 5 timescales and trendlines

obtained by linear regression for model 1-1 predictions between the years 1972-2100

on Muğla location. This location was considered at this step since modified Mann-

Kendall test results pointed a significant change in trends for all timescales. The

negative slope in SPI signifies increase in drought as negative SPI values mean dry

conditions. Thus, these plots indicate increasing drought condition for Muğla.

Further, the magnitude of trendline slope increases in negative direction as timescale

increases. This can be interpreted as the drought increase is not limited at

meteorological scale.

The significance of dry and wet periods within the evaluation period is more evident

with increasing time duration from 1-month to 12 months in SPI values (Figure 15).

Fluctuations in SPI with wet and dry values are reduced and they become more

compact. For example, with the longer duration the drought condition changes its

phase (level) from meteorological to agricultural drought and finally it may reach to

the level of hydrological drought. If the drought with its magnitude and duration is

persistent, it may be referred as agricultural to hydrological drought (SPI 9 and 12).

44

Figure 15. SPI values for Muğla station according to model 1-1 predictions (1972-

2100). The linear regression equation is on top-right hand corner.

y = -0.0004x + 0.3395

-5

-4

-3

-2

-1

0

1

2

3

4

5197

2197

6198

0198

4198

8199

2199

7200

1200

5200

9201

3201

7202

2202

6203

0203

4203

8204

2204

7205

1205

5205

9206

3206

7207

2207

6208

0208

4208

8209

2209

7

a) SPI 1 for Muğla

SPI1 Linear Regression Line

y = -0.0006x + 0.4416

-5

-4

-3

-2

-1

0

1

2

3

4

5

197

2197

6198

0198

4198

8199

2199

7200

1200

5200

9201

3201

7202

2202

6203

0203

4203

8204

2204

7205

1205

5205

9206

3206

7207

2207

6208

0208

4208

8209

2209

7

b) SPI 3 for Muğla


45


y = -0.0007x + 0.5322

-5

-4

-3

-2

-1

0

1

2

3

4

5

197

2197

6198

0198

4198

8199

2199

7200

1200

5200

9201

3201

7202

2202

6203

0203

4203

8204

2204

7205

1205

5205

9206

3206

7207

2207

6208

0208

4208

8209

2209

7

c) SPI 6 for Muğla


y = -0.0008x + 0.6518

-5

-4

-3

-2

-1

0

1

2

3

4

5

197

2197

6198

0198

4198

8199

2199

7200

1200

5200

9201

3201

7202

2202

6203

0203

4203

8204

2204

7205

1205

5205

9206

3206

7207

2207

6208

0208

4208

8209

2209

7

d) SPI 9 for Muğla


46


3.3. Modeled trend analysis of SPI-3, -12, and annual precipitation at selected

stations

Table 10 demonstrates the trend analysis results for 5 stations from different parts of

Mediterranean climate region of Turkey. This table indicates the consistency and

divergency of model predictions.

There is no significant trend in drought and annual rainfall according to four of the

models on Adana location. The forcing effect of Global Climate Models can be

inferred since three of these four models (2-1, 2-2, 2-3) are forced by same GCM

(CNRM-CERFACS-CNRM-CM5). Model 4-2 predicted no significant change at

trend for this location.

y = -0.0008x + 0.6219

-5

-4

-3

-2

-1

0

1

2

3

4

5197

2197

6198

0198

4198

8199

2199

7200

1200

5200

9201

3201

7202

2202

6203

0203

4203

8204

2204

7205

1205

5205

9206

3206

7207

2207

6208

0208

4208

8209

2209

7

e) SPI 12 for Muğla

SPI Linear Regression Line

47

Almost all models predicted an increase in drought and decrease in annual rainfall

for Antalya and Muğla locations while the analysis results are divergent for other

three locations. Only model 4-2 diverged from rest of the models since no significant

trend was detected for Muğla.

Balıkesir is the location in which relatively wetter conditions are expected. The

models did not predict an increasing drought for both 3-monthly and 12-monthly

scales except 3 of them (1-2, 3-1, 4-1). Further, model 3-2 predicted an increase in

SPI 12 that signifies wetter conditions.

The increase in drought at İzmir location is limited at 3 monthly scale for model 1-1,

1-3 and 2-2. However, models 1-2, 1-4, 2-3, 3-1, 3-2, 3-3 and 4-1 predicted a negative

slope at SPI trendline for not only 3-monthly but also 12-monthly scale. Lastly, a

decreasing annual rainfall accompany the increasing drought according to half of the

models. Model 4-2 predicted wetter conditions unlike the other 11 models.

Table 10 also enables detecting the forcing effect of Regional Climate Models. This

case is particularly obvious at Balıkesir and İzmir locations. Model 1-2 (forced by

CCLM4-8-17 RCM) predicted drier conditions for Balıkesir in contrast to other three

models which are forced by same GCM: ICHEC-EC-EARTH. IPSL-IPSL-CM5A-

MR GCM is also susceptible to RCM effect. The results are totally different on 4 of

5 locations for 2 models: 4-1 and 4-2.

The prediction divergence of the models on İzmir is similar to Balıkesir. Models 1-

2, 1-4, 2-3, 3-1, 3-2, 3-3 and 4-1 predicted drier conditions on İzmir location

considering SPI 3 and SPI 12 trends together. The forcing effect of RCA4 RCM is

significant since slope of all predictions forced by this RCM are at the same direction

no matter which GCM is forcing. Model 4-2 predicted wetter conditions in contrast

to other 11 models.

48

Table 10. Trend change table for 5 stations. ↘ denotes negative slope, ↔ denotes

no significant trend, ↗ denotes positive slope.

Locati

on

Analysed

trendlines

Model No

1-1 1-2 1-3 1-4 2-1 2-2 2-3 3-1 3-2 3-3 4-1 4-2

Adan

a

SPI 3 ↘ ↘ ↘ ↘ ↔ ↔ ↘ ↘ ↘ ↘ ↘ ↔

SPI 12 ↘ ↘ ↘ ↔ ↔ ↔ ↔ ↘ ↘ ↘ ↘ ↔

Annual ↘ ↘ ↘ ↘ ↔ ↔ ↔ ↘ ↘ ↘ ↘ ↔

An

taly

a

SPI 3 ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘

SPI 12 ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘

Annual ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘

Bal

ıkes

ir SPI 3 ↔ ↘ ↔ ↘ ↔ ↔ ↔ ↘ ↔ ↔ ↘ ↔

SPI 12 ↔ ↘ ↔ ↔ ↔ ↔ ↔ ↘ ↗ ↔ ↘ ↔

Annual ↔ ↘ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↔ ↘ ↔

İzm

ir

SPI 3 ↘ ↘ ↘ ↘ ↔ ↘ ↘ ↘ ↘ ↘ ↘ ↗

SPI 12 ↔ ↘ ↔ ↘ ↔ ↔ ↘ ↘ ↘ ↘ ↘ ↗

Annual ↔ ↘ ↔ ↘ ↔ ↔ ↘ ↘ ↔ ↘ ↘ ↗

Mu

ğla

SPI 3 ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↔

SPI 12 ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↔

Annual ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↔

Stacked line plots show the magnitude of fluctuations between years and the trends

along with the period (Figure 16). The upper lines in the y-axes show the higher

variability between the years and towards the origin the lines follow smoother curve.

These plots help us to identify the models, which indicate the largest and smallest

inter annual variability in precipitation and their general trend tendency along the

years. For example, these two features, general trend tendency and level of inter

annual variability among models appeared distinctly in Antalya and Muğla (Figure

16b and 16e respectively).

49

Figure 16. Annual precipitation for all models with stacked lines

197

2

197

7

198

2

198

7

199

2

199

7

200

2

200

7

201

2

201

7

202

2

202

7

203

2

203

7

204

2

204

7

205

2

205

7

206

2

206

7

207

2

207

7

208

2

208

7

209

2

209

7

Sta

cked lin

es f

or

annual ra

infa

ll

a) Adana

197

2

197

7

198

2

198

7

199

2

199

7

200

2

200

7

201

2

201

7

202

2

202

7

203

2

203

7

204

2

204

7

205

2

205

7

206

2

206

7

207

2

207

7

208

2

208

7

209

2

209

7

Sta

cked lin

es f

or

annual ra

infa

ll

b) Antalya

50


197

2

197

7

198

2

198

7

199

2

199

7

200

2

200

7

201

2

201

7

202

2

202

7

203

2

203

7

204

2

204

7

205

2

205

7

206

2

206

7

207

2

207

7

208

2

208

7

209

2

209

7

Sta

cked lin

es f

or

annual ra

infa

ll

c) Balıkesir197

2

197

7

198

2

198

7

199

2

199

7

200

2

200

7

201

2

201

7

202

2

202

7

203

2

203

7

204

2

204

7

205

2

205

7

206

2

206

7

207

2

207

7

208

2

208

7

209

2

209

7

Sta

cked lin

es f

or

annual ra

infa

ll d) İzmir

51

Figure16. (cont’d)

197

2

197

7

198

2

198

7

199

2

199

7

200

2

200

7

201

2

201

7

202

2

202

7

203

2

203

7

204

2

204

7

205

2

205

7

206

2

206

7

207

2

207

7

208

2

208

7

209

2

209

7

Sta

cked lin

es f

or

annual ra

infa

ll e) Muğla

52

3.4. The Slope of Trend Analysis of SPI and Annual Precipitation From

Ensemble Model Results

Following the methodology and methods presented in previous chapter, future

changes in drought and rainfall were analyzed. The corresponding data of 12

CORDEX models was extracted for 60 locations -which have the same locations with

the observation stations of the counties in the study area. SPI indices were analyzed

for 1,3,6,9 and 12 monthly timescales based on the predicted monthly precipitation

values between 1972-2100. Afterwards the trends of SPI indices and annual rainfall

for each location were tested in terms of 5% significance level. Lastly, the trend slope

of each SPI and rainfall time series which referred to a significant change according

to modified Mann-Kendall test was obtained with linear regression. The slope values

for each model and location were also presented in Appendix B.

Figure 17 The study area with province names

53

In the mapping of the trend results, only the provinces covered in the study area were

visualized (Figure 17). A red or blue circle was fixed up depending on slope direction.

No circle was fixed up for the locations in which the significant trend was not

detected. Thus, all the circles in the maps were fixed up for only the locations where

a significant trend was detected and the results were interpreted noticing this

situation.

SPI 1

Trend analysis of successive SPI 1 values is important to detect changes in short term

drought as well as it is susceptible to seasonality effects. On the other hand,

examination of trends for this timescale is necessary to observe any tendency to

meteorological drought. Following paragraphs interpret the outcomes of trend

analysis results for SPI 1 showed in Figure 18. The legend indicated in Figure 18a is

valid for all SPI trend maps in the thesis.

Model 1-1 predicted negative slope in SPI 1 trendline for all the locations which

modified Mann-Kendall test detected significant change (Figure 18a). The

magnitudes of slopes were relatively greater on the points which are located in Muğla

and Adana provinces and western Antalya as well. Another model which is forced by

same GCM (model 1-2) predicted similar results to model 1-1 (Figure 18b). However,

Denizli is the only province which a significant trend was not detected for any

location according to model 1-1 while model 1-2 indicated no significant trend for

the locations in only Bursa province (Figures 18a and 18b).

Though both of the other two models (1-3 and 1-4) from ICHEC-EC-EARTH GCM

predicted only negative slope in SPI 1 trendlines, the geographical distribution of

significant changes is limited to a smaller area comparing to models 1-1 and 1-2

(Figures 18a, 18b, 18c and 18d). Modified Mann-Kendall test results depending on

model 1-3 predictions revealed no significant trend in SPI 1 values for Bursa,

54

Balıkesir and Osmaniye provinces and eastern Antalya (Figure 18c). Though the

trend result map of model 1-4 is like model 1-3 result map in terms of the locations

with highest drought increase trend, model 1-4 also projected negative slope in SPI 1

trendline for two provinces from northern part of study area -Balıkesir and Bursa

(Figure 18d).

The expected drought increase is considerably limited to a region according to models

forced by CNRM-CERFACS-CNRM-CM5 GCM (Figures 18e, 18f and 18g). Model

2-1 predicted negative slope for locations extending from western Antalya to İzmir

(Figure 18e). Still, there are three locations in which a drought increase was expected

from eastern part of study area -at least for one-monthly timescale. This model

predicted significant trend in SPI 1 values for none of the locations from Çanakkale,

Balıkesir and Adana provinces. On the other hand, there is one location in which

positive slope was detected from each of two provinces: Bursa and Mersin.

Model 2-2 predicted greater slope magnitude for the locations in which model 2-1

projected negative slope (Figure 18f). In addition to this, the location in which

drought decrease was expected from Bursa province is same with model 2-1, but

slope magnitude is smaller. Thus, it can be interpreted that model 2-2 predicted drier

conditions than model 2-1 predicted. Balıkesir and Adana are two common provinces

in which both models predicted no significant trend.

The area in which highest drought increase trends are likely to occur according to

model 2-3 is similar to model 2-1 and 2-2 (Figures 18e, 18f and 18g). This area

includes locations from Muğla province and western part of Antalya. However, this

model predicted negative slope in SPI 1 trendline for locations from Adana province

whereas other two models from same GCM did not predict significant trend. In

addition to this, no significant trend was detected for any location in Aydın province

as distinct from model 2-1 and 2-2 (Figures 18e, 18f, and 18g). The divergence of

these trend results for SPI 1 indicated the effect of RCA4 RCM which forces model

2-3 as well as a high probability of drought increase for certain parts of study area.

55

Model 3-1 predicted negative slope in SPI 1 trendline for locations from all provinces

covered in study area (Figure 18h). Particularly, the slope magnitudes are greater at

some locations from Muğla, Denizli, Burdur and Antalya provinces. The

geographical distribution of negative slope is relatively limited according to model

3-2 and the magnitudes are smaller (Figure 18i). This model did not predict any

significant trend for locations from Bursa province and northern parts of Çanakkale

and Balıkesir as well.

The drought increase trends are highest around Adana and Hatay provinces according

to model 3-3 (Figure 18j). There is no location in which model 3-3 predicted

significant trend from Aydın province like model 2-3 which is forced by same RCM:

RCA4.

Model 4-1 diverged from rest of the models owing to greater slope magnitudes that

all indicated a significant negative trend for SPI 1 (Figure 18k). Though, the highest

drought increase trends are around two regions (Muğla-Antalya and Adana-Hatay)

which are partially common with rest of the models (Figure 18). This model also

diverged from the other model couplings which are forced by same RCM (RCA4) in

terms of the wideness of significant trend locations (Figure 18d, 18g, 18j and 18k).

The other model 4-2 which is forced by same GCM (IPSL-IPSL-CM5A-MR) is the

only model that predicted positive slope in SPI 1 trendline for many locations (Figure

18l). Aegean coast -from Aydın to Çanakkale- is the part where drought decrease

expectations at its highest level. On the other side, negative slope was obtained in SPI

1 trendline for some locations from southern part of the study area which are common

with rest of the models (Figure 18).

56

Figure 18 The geographical distribution of trends for SPI 1. The legend is valid for

rest of the SPI trend maps

57


58


59


60


61


62


Considering the effect of GCMs by comparing the model couplings from same RCM

is also another aim of this study. There are three RCMs which coupled with different

GCMs: CCLM4-8-17, RACMO22E and RCA4 (Table 2). Model 1-2, 2-2 and 3-1 are

the models which are forced by CCLM4-8-17 RCM but different GCMs respectively:

ICHEC-EC-EARTH, CNRM-CERFACS-CNRM-CM5 and MOHC-HadGEM2-ES.

The number of significant trend detected locations according to model 2-2 which is

forced by CNRM-CERFACS-CNRM-CM5 GCM is less than model 1-2 and 3-1

detections (Figures 18b, 18f and 18h). The expected drought increase area is

relatively limited for model 2-3 in comparison to other models forced by RCA4 RCM

(Figures 18d, 18g, 18j, and 18k). On the other hand, the trend analysis maps of models

1-2 and 3-1 (forced by CCLM4-8-17 RCM) are similar to each other (Figure 18b and

18h). This similarity is also valid between models 1-3 and 3-2 (both forced by

RACMO22E RCM) and between models 1-4 and 3-3 (both forced by RCA4 RCM).

These matchings highlight the distinctness of CNRM-CERFACS-CNRM-CM5

GCM from other two GCMs (ICHEC-EC-EARTH and MOHC-HadGEM2-ES)

63

which were considered in this comparison. IPSL-IPSL-CM5A-MR is also a more

impact GCM since model 4-1 projects a quite wider drought increase area with

greater magnitudes than models 1-4, 2-3 and 3-3 predict which are forced by same

RCM: RCA4 (Figures 18d, 18g, 18j and 18k).

SPI 3

Significant trend in successive SPI 3 values can be interpreted a first step to

agricultural drought. In this part the trends in SPI 3 were interpreted comparing the

models and SPI 1 trend results.

Considering model 1-1, two different behavior on SPI trends occurred for the

locations in which a significant trend was detected for SPI 1 values (Figure 18a and

19a). Slope magnitudes became greater at this timescale for most of the locations in

which a negative slope was detected in SPI 1 trendline whereas significant trend did

not exist anymore for some of them. On the other hand, positive slope in SPI 3

trendline was detected for four locations from different provinces: Bursa, Balıkesir,

Antalya and Denizli. Still, the highest drought increase expectations according to SPI

3 trends are on the same area with SPI 1: Muğla, Antalya and Adana. This fact

indicated a tendency to agricultural drought increase.

Though significant trend disappeared for a few locations from northern Aegean coast,

rest of the locations in which a negative slope was estimated for SPI 1 signified a

greater magnitude for SPI 3 for model 1-2 (Figure 18b and 19b). This projection

pretty much fitted model 1-1 projection in terms of tendency to agricultural drought.

However, the drought increasing expectations became evident for a region from

western Antalya to İzmir and Mersin tend to live drier conditions rather than Adana

as a distinction from model 1-1 (Figures 19a and 19b).

64

There are new locations in which a significant trend (negative slope) for SPI 3 was

detected from southern part of the study area according to model 1-3 (Figure 19c).

Two locations from Manisa did not signify a significant trend at this timescale

anymore. The drought increase expectation aggregated around Muğla and western

Antalya according to this model considering the increasing slope magnitudes on

negative direction. Yet, the significant trend detected locations for SPI 3 are relatively

limited for model 1-3 in comparison to other models from same GCM (Figures 19a,

19b, 19c and 19d). This variation was also valid for SPI 1 trends (Figure 18a, Figure

18b, Figure 18c and Figure 18d).

Model 1-4 (Figure 19d) predicted greater slope magnitudes on negative direction for

SPI 3. The drought increase expectation is most likely for an area extending from

western Antalya to İzmir. Besides, there are new locations in which a significant trend

was detected from Mersin province.

The significant trend detected locations are even more limited for SPI 3 than SPI 1

according to model 2-1 (Figure 19e). However, the slope magnitudes increased on

both negative and positive directions comparing to SPI 1 (Figure 18e). This fact is

also valid for model 2-2 (Figure 18f and 19f). There is also one new location with

positive slope from Bursa province.

There is no location with significant trend on negative direction for SPI 3 from İzmir,

Denizli and Osmaniye provinces as distinct from SPI 1 for model 2-3 (Figure 18g

and 19g). This model predicted wetter conditions for southern Marmara considering

4 locations with positive slope from Çanakkale, Balıkesir and Bursa provinces.

Though, the negative slope magnitude increased for some locations from Muğla and

Gaziantep.

65

Figure 19. The geographical distribution of trends for SPI 3

66


67


68


69


70


71

Model 3-1 (Figure 19h) predicted negative slope in SPI 3 trendlines for all provinces

except Bursa. The locations with greatest slope magnitudes are from Muğla, western

Antalya, Denizli, Hatay and Gaziantep at this timescale. The drought tendency is not

limited at meteorological level for almost whole study area according to this model.

The drought increasing expected locations shifted southernly according to model 3-

2 considering the differences between SPI 1 and SPI 3 (Figure 18f and 19f). Six

locations from Manisa, İzmir and Aegean coast of Balıkesir did not signify a negative

SPI slope at three-months timescale as well as new locations with negative slope

came out from Antalya, Mersin, Adana and Hatay provinces. Besides, a positive slope

was detected for four locations from southern Marmara.

The magnitudes of slopes in SPI 3 trendlines increased overall for model 3-3

projections comparing to SPI 1 (Figures 18j and 19j). Additionally, Manisa and

Denizli cover locations with negative slopes at this timescale. In general, the slope

magnitudes are close to each other for southern part of study area (along

Mediterranean coastline).

The geographical distributions of SPI 3 trend results are almost same with SPI 1 trend

results for model 4-1 and 4-2 (Figures 18k, 18l, 19k and 19l). However, the slopes

sharpened in both directions like rest of the models (Figure 18 and 19).

The regions in which drought expectations aggregated according to SPI 1 trendlines

(Muğla, western Antalya and Adana) keep this characteristic at SPI 3 as well (Figure

18 and 19). This fact can be interpreted as a transition from meteorological drought

to agricultural drought. However, the disappeared drought increasing signs for some

locations should be noticed (particularly for northern parts of the study area).

72

SPI 6

Considering the trendlines for SPI 6, the results of all models tend to predict drier

conditions for almost same areas with SPI 3 (Figures 19 and 20). However, all models

indicated greater slope magnitudes for both negative and positive directions. These

facts signify a tendency to hydrological drought for the locations which agricultural

drought increase likely to occur.

Muğla and western Antalya is the region which all models project drought increases

at 6-monthly timescale (Figure 20). Model 4-2 and three models (2-1, 2-2, 2-3) forced

by CNRM-CERFACS-CNRM-CM5 GCM diverged from rest of the models since

they do not predict drier conditions for Adana province as much as the others

predicted (Figures 20e, 20f, 20g and 20l). The northern parts of study area (Southern

Marmara) are likely not to live drier conditions at this timescale according to most of

the models. However, İzmir is the province which model projections did not fit each

other widely. Model 1-1, 1-2, 1-3, 1-4, 3-1, 3-2, 3-3 and 4-1 predicted a negative

slope in SPI 6 trendlines for more than one location from this province (Figure 20).

On the other hand, models 2-1 and 2-2 predicted drier conditions for only one location

from İzmir while models 2-3 and 4-2 did not predict negative slope for any location

in İzmir (Figures 20e, 20f, 20g and 20l).

73


74


75


76


77


78


79

SPI 9

Though the trend result maps for SPI 9 resembles SPI 6 result maps, there are some

locations which slope direction changed at this timescale (Figures 20 and 21).

There are seven new locations in which a significant trend was detected for SPI 9

according to model 1-2 (Figure 21b). All of them are from northern part of study area

-from Bursa, Balıkesir and Çanakkale provinces. One location from Bursa is likely

to live wetter conditions at this timescale whereas the rest are drier. The rest of the

trend result maps for SPI 9 did not change a lot from SPI 6 compared to model 1-1

and 1-2 (Figures 20a, 20b, 21a and 21b). Still, the sharpening slopes should be noticed

for these models.


80


81


82


83


84


85


SPI 12

Evaluating trends in SPI 12 values is important since dryness for 12 months indicate

a serious tendency to hydrological drought. The geographical distribution of regions

which expected drier and wetter conditions aggregated for SPI 12 is same with SPI

9, SPI 6 and even SPI 3 (from Figure 19 to 22). Muğla and western Antalya is still

the most vulnerable region to drought increase whereas northern part of study area

(southern Marmara) is likely to live wetter conditions (Figure 22). However, trends

for different timescales are not persistent at some locations. Four locations from

Bursa and Çanakkale provinces which signaled a significant trend at negative slope

for SPI 9 disappeared at 12-monthly timescale according to model 1-2 (Figures 21b

and 22b).

86


87


88


89


90


91


92

İzmir should be separately discussed in terms of continuity of trends for proceeding

timescales. Models 1-1 and 3-2 predicted negative slope at fewer locations from İzmir

for SPI 12 than SPI 9 and model 2-1 vice versa (Figures 21a, 21i, 21e, 22a, 22i and

22e). This fact indicated a discrepancy between models for certain locations.

Annual Precipitation

Detecting trends in annual precipitation is another part of this study. Relations

between drought and rainfall amount can be examined owing to this analysis. In this

part, the geographical distribution of annual rainfall trends was discussed with trends

in SPI.

Model 1-1 (Figure 23a) predicted a decrease in annual rainfall for the locations which

are sensitive to drought increases according to SPI trends. Muğla and western Antalya

is the most sensitive region while a decreasing trend is valid along with whole

Mediterranean coast. However, neither rainfall decrease accompanied drought

increase at every location nor rainfall increase accompanied drought decrease at

every location (Figures 22a and 23a). Model 1-2 is like to model 1-1 in terms of

drought and rainfall trend consistency (Figures 22b and 23b). Nonetheless, annual

rainfall decrease expectations is more common according to model 1-2 projection.

Not only Mediterranean coast but also Aegean region is likely to receive less annual

precipitation.

Model 1-3 (Figure 23c) predicted annual rainfall decrease for a smaller area than most

of the models predicted. Muğla and western Antalya is the region which drought

increases almost followed rainfall decreases. The locations which are most likely to

receive less annual rainfall according to model 1-4 is same with rest of the models

(Figure 23). On the other hand, the vulnerable locations across entire study area like

model 1-2 (Figures 23b and 23d).

93

The parts of the study area which signaled trends in annual rainfall are almost same

for models 2-1, 2-2 and 2-3 which are forced by CNRM-CERFACS-CNRM-CM5

GCM (Figures 23e, 23f and 23g). These models predicted rainfall increase for at least

one location from Bursa province. In addition to this, few locations from Balıkesir

and Çanakkale are likely to receive more annual rainfall according to model 2-3

(Figure 23g). Model 3-1 predicted decrease in annual rainfall across the entire study

area like model 1-2, 1-4 and 4-1 (Figure 23). Model 1-2 is forced by same RCM

(CCLM4-8-17) with model 3-1 while models 1-4 and 4-1 are forced by RCA4 RCM.

Mediterranean coast is likely to receive less annual precipitation rather than Aegean

coast according to models 3-2 and 3-3 (Figures 23i and 23j). Additionally, drought

increase accompanied precipitation decrease in most of the locations for model

predictions discussed in this paragraph.

94

Figure 23. The geographical distribution of trends for annual rainfall

95


96


97


98


99


100


Model 4-1 predicted rainfall decrease for almost all locations like its predictions on

drought trends (Figure 23k). Model 4-2 (Figure 23l) diverged from rest of the models

considering its projections which signaled rainfall increase almost entire Aegean

coast and some other parts like eastern Antalya.

101

Changes in the Frequency of Drought Between the Mid and End of the

Century

Changes in intensity of drought were analyzed and mapped in this step of the study.

The projection period is divided into two periods: 1972-2050 and 2051-2100. Then,

the frequency of occurrence of three drought severity class (moderate, severe and

extreme) depending on SPI values was calculated through divided periods for 3 and

12 monthly scales. The proportion of frequencies was obtained at the final step as

shown in Eq. 17:

r =

o2

n2o1

n1

(17)

where, r is the ratio of frequencies for each drought severity class, o1 is number of

occurrences for a severity class within 1972-2050 period, o2 is the same as within

2051-2100 period, n1 and n2 are number of total months within each period.

In the mapping of the ratios, a base map prepared with provincial borders was

practiced. Though some provincial borders do not entirely fit to the study area since

they extend inner parts of the country, this mismatch was partly eliminated owing to

the implied interpolation method: Natural neighbor. This method was firstly

introduced by Sibson (1981), and ArcGIS software was used in practice of this study.

The limits at legends of the maps were constituted with the highest value of changes,

concerning demonstration of the divergence between models. The result maps were

presented in Figure 24.

The most significant result of this analysis is the frequency increase in the second

half of the century for severe and extreme droughts consistently for almost all models

and timescales (Figure 24). To clarify, the drought gets more often as the intensity

gets bigger. More frequent droughts are more possible for southern provinces, either.

102

The geographical distribution of ratios is consistent with trend analysis results.

Southern parts of study area are likely to occur more intense drought. Besides, the

geographical distribution of the ratios of frequency is quite homogeneous for SPI 3

compared to SPI 12 for all models. The highest ratios aggregate in certain parts of

study area (especially southwestern parts) considering changes in drought frequency

according to SPI 12 values. For example, the ratio exceeds 62 times for model 1-4

(Figure 24) under extreme drought severity condition considering SPI 12.

103

Fig

ure

24.

Changes

in t

he

freq

uency o

f dro

ug

ht. ‘

Mo

d’,

‘se

v’

and ‘

ext’

ind

icat

e m

oder

atel

y,

sever

ely

and e

xtr

em

ely

dry

cla

sses

res

pec

tively

.

104

Fig

ure

24.

(co

nt’

d)

105

Fig

ure

24.

(co

nt’

d)

106

Fig

ure

24

. (c

ont’

d)

107

Fig

ure

24.

(co

nt’

d)

108

Fig

ure

24.

(co

nt’

d)

109

Fig

ure

24.

(co

nt’

d)

110

Fig

ure

24.

(co

nt’

d)

111

Fig

ure

24.

(co

nt’

d)

112

Fig

ure

24.

(co

nt’

d)

113

Fig

ure

24.

(co

nt’

d)

114

Fig

ure

24.

(co

nt’

d)

115

Areal Extension of Drought within the Period

The plots in Figure 25 show the areal change of drought based on SPI severity

classes. The percentage of the locations at each month which is in relevant SPI value

interval for each class was calculated. Afterwards, the monthly averages were

obtained for 4 periods: 1971-2005, 2006-2040, 2041-2075, 2076-2099. SPI 3 values

were considered at this evaluation. The area under dry conditions is no more than

30% for the driest model (4-1), yet, the increase is related to intensity as the plots

indicated. For instance, extremely dry area does not exceed 6 % whereas the sharpest

increase is the case for this severity class.

Figure 25. The percentage of the stations under dry conditions

0

2

4

6

8

10

12

14

16

18

1-1 1-2 1-3 1-4 2-1 2-2 2-3 3-1 3-2 3-3 4-1 4-2

Per

cen

tage

of

the

stat

ion

s

a) Moderately dry

1972-2005 2006-2040 2041-2075 2076-2099

116


0

2

4

6

8

10

12

14

16

18

1-1 1-2 1-3 1-4 2-1 2-2 2-3 3-1 3-2 3-3 4-1 4-2

Per

cen

tage

of

the

stat

ion

s

b) Severely dry

1972-2005 2006-2040 2041-2075 2076-2099

0

2

4

6

8

10

12

14

16

18

1-1 1-2 1-3 1-4 2-1 2-2 2-3 3-1 3-2 3-3 4-1 4-2

Per

cen

tage

of

the

stat

ion

s

c) Extremely dry

1972-2005 2006-2040 2041-2075 2076-2099

117

Figure 26. The average percentage of the stations under dry conditions for all models

Figure 26 indicates the percentages of stations under dry conditions by averaging

percentiles of 12 models. Though the percentages raised for all dryness classes, the

increase rates are diverse. While moderate dry locations raised from 7% to 11%,

percentages of severe and extreme dry locations raised almost 2 times (from 3% to

6% and from 1.5% to 4% respectively)

0

2

4

6

8

10

12

197

2-2

005

200

6-2

040

204

1-2

075

207

6-2

099

197

2-2

005

200

6-2

040

204

1-2

075

207

6-2

099

197

2-2

005

200

6-2

040

204

1-2

075

207

6-2

099

Moderate Severe Extreme

Perc

enta

ge o

f th

e locations

Model Mean

118

119

CHAPTER 4

SUMMARY, CONCLUSION AND RECOMMENDATIONS

In this thesis, Ensemble modeling approach was used to make predictions for drought

in the future of Turkey’s Mediterranean climate region. Observation and model

precipitation data were compared to measure the performance of each model within

the ensemble system. Afterwards, trends in drought depending on Standardized

Precipitation Index values and annual rainfall amounts were investigated.

Geographical distribution of trends was mapped. Frequency of different drought

severities was compared for two periods divided at half of 21st century. Areal

extension of drought severities was studied for 4 periods as well.

Models 1-4, 2-3 and 3-3 are distinguished from rest of the models with their high

consistency with observation data. It should be noticed that these models used the

same RCM (RCA4) but initiated with different GCMs. Additionally, RCM show

more impact on predictions considering the uncertainties that come from GCM and

RCM. Model 1-1 is also distinguished owing to its well correlation with observation

data considering daily means. Though, this model diverged from observation at 2 of

5 locations when annual rainfall trends were compared., Model 4-1 is another model

from RCA4 RCM that partially fitted observation though its underestimation remains

for entire year.

In case of projections, south of the study area (especially Muğla province and western

part of Antalya province) is the most sensitive region to drought in future according

to all model predictions. This result is consistent with other studies which

investigated drought trends in Turkey (Topcu et al., 2010; Sen et al., 2012). Hence,

120

the closeness of the results from all models for these regions indicated the consistency

of the models and the importance of drought signals. On the other hand, the variation

of the trend results of models for the rest of the area -especially along the Aegean

region- causes the necessity for evaluating the RCM projections more in detail in this

area. Most of the models project either no significant trend or wetter conditions for

locations from Çanakkale, Balıkesir and Bursa. Besides, the Marmara coast of these

provinces is likely to receive more precipitation.

Another significant result that can be inferred from both result tables and maps is the

increasing drought tendency for larger timescales. This means that the drier

conditions may not be limited to meteorological or agricultural scale. Increase in

hydrological drought is probable for most of the study area. Therefore, the

measurements that will be held to manage the effects of climate change should be

designed taking into account this projection outcome at this region.

The analyses revealed drought increase with decreasing trend at annual total

precipitation for most of the locations. Increase in drought may have a positive

correlation with decrease in annual precipitation in the region. However, the more

detailed drought and precipitation assessments considering the correlation between

these should be performed.

The increase in the frequency of different drought types based on SPI classification

is related to severity of class according to most of the models. This analysis result is

common with findings of Leng et al. (2015) on China. Areal extension proportions

of drought classes also consist with this result. Severely and extremely dry areas

extend more than moderately dry areas proportionally.

This thesis was dedicated to Mediterranean climate region of Turkey. The drought

analysis can be extended to entire Turkey. Other drought indices can also be implied.

Changes in temperature should also be studied to clarify the effect of climate change.

The relation between trends in drought and changes in seasonal distribution of

121

precipitation can be analyzed. The effects of climate change should be assessed via

analyzing the phenomena at larger scale as well. Poleward shift of Hadley cell is one

of these phenomena. Evaluating the trends in meteorological parameters on the belt

around 30° latitude will be useful to determine any significant change.

122

REFERENCES

Abramowitz, M. and Stegun, A. (eds). (1965). Handbook of Mathematical Formulas,

Graphs, and Mathematical Tables. New York.: Dover Publications.

Cook, B. I., Anchukaitis, K. J., Touchan, R., Meko, D. M., & Cook, E. R. (2016).

Spatiotemporal drought variability in the Mediterranean over the last 900 years.

Journal of Geophysical Research: Atmospheres, 121(5), 2060–2074.

https://doi.org/10.1002/2015JD023929

Dabanlı, İ., Mishra, A. K., & Şen, Z. (2017). Long-term spatio-temporal drought

variability in Turkey. Journal of Hydrology, 552, 779–792.

https://doi.org/10.1016/j.jhydrol.2017.07.038

Easterling, D. R., Meehl, G. A., Parmesan, C., Changnon, S. A., & Karl, T. R., and

Mearns, L. O. (2000). Climate extremes: Observations, modeling, and impacts.

Science, 289(2068–2074).

Edwards, D. C., & McKee, T. B. (1997). Characteristics of 20th century drought in

United State at multiple time scales. Climatology Report No. 97-2, (634), 1–

155. http://weather.uwyo.edu/upperair/sounding.html

Giorgi, F. (2006). Climate change hot-spots. Geophysical Research Letters, 33(8), 1–

4. https://doi.org/10.1029/2006GL025734

Giorgi, F., & Lionello, P. (2008). Climate change projections for the Mediterranean

region. Global and Planetary Change, 63(2–3), 90–104.

https://doi.org/10.1016/j.gloplacha.2007.09.005

Gumus, V., & Algin, H. M. (2017). Meteorological and hydrological drought analysis

of the Seyhan−Ceyhan River Basins, Turkey. Meteorological Applications,

24(1), 62–73. https://doi.org/10.1002/met.1605

Guttman, N. B. (1999). Accepting the Standardized Pre- cipitation Index: A

calculation algorithm. J. Amer. Water Resour. Assoc, 35, 311–322.

Hamed, K. H., & Ramachandra Rao, A. (1998). A modified Mann-Kendall trend test

for autocorrelated data. Journal of Hydrology, 204(1–4), 182–196.

https://doi.org/10.1016/S0022-1694(97)00125-X

Heffernan, O. (2016). The mystery of the expanding tropics. Nature, 530(7588), 20–

22. https://doi.org/10.1038/530020a

https://doi.org/10.1002/2015JD023929

https://doi.org/10.1016/j.jhydrol.2017.07.038

http://weather.uwyo.edu/upperair/sounding.html

https://doi.org/10.1029/2006GL025734


https://doi.org/10.1002/met.1605

https://doi.org/10.1016/S0022-1694(97)00125-X

https://doi.org/10.1038/530020a

123

Kara, F., Yucel, I., & Akyurek, Z. (2016). Climate change impacts on extreme

precipitation of water supply area in Istanbul: use of ensemble climate

modelling and geo-statistical downscaling. Hydrological Sciences Journal,

61(14), 2481–2495. https://doi.org/10.1080/02626667.2015.1133911

Kendall, M. G. (1975). Rank Correlation Methods.

Keyantash, J., & Dracup, J. A. (2002). The quantification of drought: An evaluation

of drought indices. Bulletin of the American Meteorological Society, 83(8),

1167–1180. https://doi.org/10.1175/1520-

0477(2002)083<1191:TQODAE>2.3.CO;2

Kottek, M., Grieser, J., Beck, C., Rudolf, B., & Rubel, F. (2006). World Map of the

Köppen-Geiger climate classification updated. Meteorologische Zeitschrift,

15(3), 259–263. https://doi.org/10.1127/0941-2948/2006/0130

Lee, S. H., Yoo, S. H., Choi, J. Y., & Bae, S. (2017). Assessment of the impact of

climate change on drought characteristics in the Hwanghae Plain, North Korea

using time series SPI and SPEI: 1981–2100. Water (Switzerland), 9(8).

https://doi.org/10.3390/w9080579

Leng, G., Tang, Q., & Rayburg, S. (2015). Climate change impacts on

meteorological, agricultural and hydrological droughts in China. Global and

Planetary Change, 126, 23–34.


Lloyd-Hughes, B., & Saunders, M. A. (2002). A drought climatology for Europe.

International Journal of Climatology, 22(13), 1571–1592.

https://doi.org/10.1002/joc.846

Mann, H. B. (1945). Non-parametric tests againist trends. Econometrica, 13, 163–

171.

Mckee, T. B., Doesken, N. J., & Kleist, J. (1993). The relationship of drought

frequency and duration to time scales. AMS 8th Conference on Applied

Climatology, (January), 179–184. https://doi.org/citeulike-article-id:10490403

Osuch, M., Romanowicz, R. J., Lawrence, D., & Wong, W. K. (2016). Trends in

projections of standardized precipitation indices in a future climate in Poland.

Hydrology and Earth System Sciences, 20(5), 1947–1969.

https://doi.org/10.5194/hess-20-1947-2016

Raleigh, C., Jordan, L., & Salehyan, I. (2008). Assessing the Impact of Climate

Change on Migration and Conflict. World, 24, 1–57.

http://siteresources.worldbank.org/EXTSOCIALDEVELOPMENT/Resources

/SDCCWorkingPaper_MigrationandConflict.pdf

https://doi.org/10.1080/02626667.2015.1133911

https://doi.org/10.1175/1520-0477(2002)083%3c1191:TQODAE%3e2.3.CO;2

https://doi.org/10.1175/1520-0477(2002)083%3c1191:TQODAE%3e2.3.CO;2

https://doi.org/10.1127/0941-2948/2006/0130

https://doi.org/10.3390/w9080579


https://doi.org/10.1002/joc.846

https://doi.org/citeulike-article-id:10490403

https://doi.org/10.5194/hess-20-1947-2016

http://siteresources.worldbank.org/EXTSOCIALDEVELOPMENT/Resources/SDCCWorkingPaper_MigrationandConflict.pdf

http://siteresources.worldbank.org/EXTSOCIALDEVELOPMENT/Resources/SDCCWorkingPaper_MigrationandConflict.pdf

124

Randall, D. A., Wood, R. A., Bony, S., Colman, R., Fichefet, T., Fyve, J., … Taylor,

K. E. (2007). Climate Models and Their Evaluation. Climate Change 2007: The

Physical Science Basis. Contribution of Working Group I to the Fourth

Assessment Report of the Intergovernmental Panel on Climate Change, 591–

662. https://doi.org/10.1016/j.cub.2007.06.045

Redmond, K. T. (2002). The depiction of drought, Bulletin of the American

Meteorological Society, 14(3), 243–260.

Reichler, T., & Kim, J. (2008). How well do coupled models simulate today’s

climate? Bulletin of the American Meteorological Society, 89(3), 303–311.

https://doi.org/10.1175/BAMS-89-3-303

Schaller, N., Mahlstein, I., Cermak, J., & Knutti, R. (2011). Analyzing precipitation

projections: A comparison of different approaches to climate model evaluation.

Journal of Geophysical Research Atmospheres, 116(10), 1–14.

https://doi.org/10.1029/2010JD014963

Sen, B., Topcu, S., Türkeş, M., Sen, B., & Warner, J. F. (2012). Projecting climate

change, drought conditions and crop productivity in Turkey. Climate Research,

52(1), 175–191. https://doi.org/10.3354/cr01074

Sensoy, S., Demircan, M., & Ulupınar, Y. Climate of Turkey.

https://www.mgm.gov.tr/files/en-US/climateofturkey.pdf

Sönmez, F. K., Kömüscü, A. Ü., Erkan, A., & Turgu, E. (2005). An analysis of spatial

and temporal dimension of drought vulnerability in Turkey using the

standardized precipitation index. Natural Hazards, 35(2), 243–264.

https://doi.org/10.1007/s11069-004-5704-7

Stagge, J. H., Rizzi, J., Tallaksen, L. M., & Stahl, K. (2015). Future meteorological

drought: projections of regional climate models for Europe, (25).

http://www.eu-

drought.org/media/default.aspx/emma/org/10859960/DROUGHT+RSPI+Tec

hnical+Report+No.+25+-

+Future+Meteorological+Drought+Projections+of+Regional+Climate.pdf

Stocker, T. F. (2014). Climate change 2013, IPCC Report – The Physical Science

Basis.

Thom, H. C. S. (1958). a Note on the Gamma Distribution. Monthly Weather Review,

86(3), 117–122. https://doi.org/10.1175/1520-

0493(1958)086<0117:ANOTGD>2.0.CO;2

https://doi.org/10.1016/j.cub.2007.06.045

https://doi.org/10.1175/BAMS-89-3-303

https://doi.org/10.1029/2010JD014963

https://doi.org/10.3354/cr01074

https://www.mgm.gov.tr/files/en-US/climateofturkey.pdf

https://doi.org/10.1007/s11069-004-5704-7

http://www.eu-drought.org/media/default.aspx/emma/org/10859960/DROUGHT+RSPI+Technical+Report+No.+25+-+Future+Meteorological+Drought+Projections+of+Regional+Climate.pdf




https://doi.org/10.1175/1520-0493(1958)086%3c0117:ANOTGD%3e2.0.CO;2

https://doi.org/10.1175/1520-0493(1958)086%3c0117:ANOTGD%3e2.0.CO;2

125

Topcu, S., Sen, B., & Türkes, M. (2010). Observed and projected changes in drought

conditions of Turkey. Options Méditerranéennes. Séries A. Mediterranean

Seminars.

Türkeş, M. (2012). Türkiye’de Gözlenen ve Öngörülen İklim Değişikliği, Kuraklık

ve Çölleşme. Ankara Üniversitesi Çevrebilimleri Dergisi, 4(2), 1–32.

Türkeș, M., Yozgatlıgil, C., Batmaz, İ., İyigün, C., E, K. K., FM, F., & Aslan, S.

(2016). Has the climate been changing in Turkey? Regional climate change

signals based on a comparative statistical analysis of two consecutive time

periods, 1950-1980 and 1981-2010 . Climate Research, 70(1), 77–93.

http://www.int-res.com/abstracts/cr/v70/n1/p77-93

WEB1: https://pmm.nasa.gov/education/water-cycle, last access: December 9, 2017.

WEB2: https://www.nasa.gov/audience/forstudents/5-8/features/nasa-knows/what-

is-climate-change-58.html, last access: December 9, 2017.

WEB3: http://cordex.org/about/what-is-regional-downscaling/, last access:

December 9, 2017.

WEB4: https://www.mgm.gov.tr/veridegerlendirme/yagis-raporu.aspx#sfU, last

access: May 23, 2018.

WEB5: https://www.usatoday.com/story/weather/2015/08/19/california-drought-

cost-27-billion-2015/32007967/, last access: January 14, 2018.

WEB6: http://www.haberturk.com/kurak-gecen-kis-elektrik-fiyatlarinda-artisa-

neden-olacak-1789971-ekonomi, last access: January 14, 2018.

WEB7: http://drought.unl.edu/DroughtBasics/TypesofDrought.aspx, last access:

January 14, 2018.

WEB8: http://vlscop.vermontlaw.edu/2016/11/16/implementing-adaptation-for-

resilient-mediterranean-climate-regions/, last access: May 23, 2018.

WEB9: http://www.eba.gov.tr/dokuman/353?icerik-

id=3798f6b4aca17c4e040f9870262671cec553ddfbde001&channel=60, last

access: January 24, 2018.

WEB10: https://portal.enes.org/data/enes-model-data/cordex/datastructure, last

access: January 24, 2018.

Wilhite, D. (1985). Understanding the Drought Phenomenon: The Role of

Definitions, (August 2013), 37–41.

http://www.int-res.com/abstracts/cr/v70/n1/p77-93

https://pmm.nasa.gov/education/water-cycle

https://www.nasa.gov/audience/forstudents/5-8/features/nasa-knows/what-is-climate-change-58.html

https://www.nasa.gov/audience/forstudents/5-8/features/nasa-knows/what-is-climate-change-58.html

http://cordex.org/about/what-is-regional-downscaling/

https://www.mgm.gov.tr/veridegerlendirme/yagis-raporu.aspx#sfU

https://www.usatoday.com/story/weather/2015/08/19/california-drought-cost-27-billion-2015/32007967/

https://www.usatoday.com/story/weather/2015/08/19/california-drought-cost-27-billion-2015/32007967/

http://www.haberturk.com/kurak-gecen-kis-elektrik-fiyatlarinda-artisa-neden-olacak-1789971-ekonomi

http://www.haberturk.com/kurak-gecen-kis-elektrik-fiyatlarinda-artisa-neden-olacak-1789971-ekonomi

http://drought.unl.edu/DroughtBasics/TypesofDrought.aspx

http://vlscop.vermontlaw.edu/2016/11/16/implementing-adaptation-for-resilient-mediterranean-climate-regions/

http://vlscop.vermontlaw.edu/2016/11/16/implementing-adaptation-for-resilient-mediterranean-climate-regions/

http://www.eba.gov.tr/dokuman/353?icerik-id=3798f6b4aca17c4e040f9870262671cec553ddfbde001&channel=60

http://www.eba.gov.tr/dokuman/353?icerik-id=3798f6b4aca17c4e040f9870262671cec553ddfbde001&channel=60

https://portal.enes.org/data/enes-model-data/cordex/datastructure

126

APPENDICES

A. MATLAB CODES

MATLAB code for Modified Mann-Kendall Test (prepared by Simone Fatichi)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%% MANN-KENDALL TEST MODIFIED by Hamed and Rao, (1998) %%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function[H,p_value]=Mann_Kendall_Modified(V,alpha)

%%%%%%%%%%%%%%%%%

%%% Performs Mann-Kendall test modified to account for autocorrelation on

the time series

%%% The null hypothesis of trend

%%% absence in the vector V is tested, against the alternative of trend.

%%% The result of the test is returned in H = 1 indicates

%%% a rejection of the null hypothesis at the alpha significance level. H

= 0 indicates

%%% a failure to reject the null hypothesis at the alpha significance

level.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%% INPUTS

%V = time series [vector]

%alpha = significance level of the test [scalar]

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%% From Matlab Help %%%%%%%%%%%%%%%

%The significance level of a test is a threshold of probability a agreed

%to before the test is conducted. A typical value of alpha is 0.05. If the

p-value of a test is less than alpha,

%the test rejects the null hypothesis. If the p-value is greater than

alpha, there is insufficient evidence

%to reject the null hypothesis.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%% OUTPUTS

https://www.mathworks.com/matlabcentral/profile/authors/1881857-simone-fatichi

127

%H = test result [1] Reject of Null Hypthesis [0] Insufficient evidence to

reject the null hypothesis

%p_value = p-value of the test

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%% From Matlab Help %%%%%%%%%%%%%%%

%The p-value of a test is the probability, under the null hypothesis, of

obtaining a value

%of the test statistic as extreme or more extreme than the value computed

from

%the sample.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%% References

%Mann, H. B. (1945), Nonparametric tests against trend, Econometrica, 13,

%245 259.

%Kendall, M. G. (1975), Rank Correlation Methods, Griffin, London.

%Hamed, K. H., and A. R. Rao (1998), A modified Mann-Kendall trend test

%for autocorrelated data, J. Hydrol., 204, 182196.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% Simone Fatichi -- [email protected]

% Copyright 2009

% $Date: 2009/10/03 $

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

V=reshape(V,length(V),1);

alpha = alpha/2; %

n=length(V);

i=0; j=0; S=0;

for i=1:n-1

for j= i+1:n

S= S + sign(V(j)-V(i));

end

end

VarSo=(n*(n-1)*(2*n+5))/18;

%%%%%%%%%%%%%%%%%%%%%%%%

ANSW = 3; %%% It depends on computational time

switch ANSW

case 1

xx=1:n;

aa=polyfit(xx,V,1);

yy=aa(1,1)*xx+aa(1,2);

V=V-yy';

case 2

[b]=Sen_Slope(V);

128

xx=1:n;

yy=b*xx+ (mean(V) -(b*n)/2);

V=V-yy';

case 3

V=detrend(V);

end

%%%%%%%%%%%%%%%%%%%%%%%%%

[V,I]=sort(V); %% I = ranks

[Acx,lags,Bounds]=autocorr(I,n-1);

%[Acx,lags]=xcov(I,I,n-1,'coeff'); %%

%Acx=Acx(n:end);

ros=Acx(2:end); %% Autocorrelation Ranks

i=0; sni=0;

for i=1:n-2

if ros(i)<= Bounds(1) andand ros(i) >= Bounds(2)

sni=sni;

else

sni=sni+(n-i)*(n-i-1)*(n-i-2)*ros(i);

end

end

nns=1+(2/(n*(n-1)*(n-2)))*sni;

VarS=VarSo*(nns);

StdS=sqrt(VarS);

if S >= 0

Z=((S-1)/StdS)*(S~=0);

else

Z=(S+1)/StdS;

end

p_value=2*(1-normcdf(abs(Z),0,1));

pz=norminv(1-alpha,0,1);

H=abs(Z)>pz; %

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%%%% Trend Magnitude ---> Sen (1968)%%

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function[b]=Sen_Slope(X)

i=0; %

n=length(X);

V= zeros(1,(n^2-n)/2);

for j=2:n

for l=1:j-1

i=i+1;

129

V(i)=(X(j)-X(l))/(j-l);

end

end

b=median(V);

end

130

MATLAB code for Standardized Precipitation Index (prepared by Taesam Lee)

%%

% Programmed by Taesam Lee, Dec.03,2009

% INRS-ETE, Quebec, Canada

function [Z]=SPI(Data,scale,nseas)

%Standardized Precipitation Index

% Input Data

% Data : Monthly Data vector not matrix (monthly or seasonal

precipitation)

% scale : 1,3,12,48

% nseas : number of season (monthly=12)

% Example

% Z=SPI(gamrnd(1,1,1000,1),3,12); 3-monthly scale,

% Notice that the rest of the months of the fist year are removed.

% eg. if scale =3, fist year data 3-12 SPI values are not estimated.

%if row vector then make coloumn vector

%if (sz==1) Data(:,1)=Data;end

erase_yr=ceil(scale/12);

% Data setting to scaled dataset

A1=[];

for is=1:scale, A1=[A1,Data(is:length(Data)-scale+is)];end

XS=sum(A1,2);

if(scale>1), XS(1:nseas*erase_yr-scale+1)=[]; end

for is=1:nseas

tind=is:nseas:length(XS);

Xn=XS(tind);

[zeroa]=find(Xn==0);

Xn_nozero=Xn;Xn_nozero(zeroa)=[];

q=length(zeroa)/length(Xn);

parm=gamfit(Xn_nozero);

Gam_xs=q+(1-q)*gamcdf(Xn,parm(1),parm(2));

Z(tind)=norminv(Gam_xs);

end

%Gamma parameter estimation and tranform

https://www.mathworks.com/matlabcentral/profile/authors/811370-taesam-lee

131

B. TREND SLOPE TABLES

Table 11. Trend slope values for model 1-1

Station SPI 1 SPI 3 SPI 6 SPI 9 SPI 12 Annual Pr.

BANDIRMA - 0.0001 0.0003 0.0004 0.0004 0.7152

AYVALIK - - 0.0001 0.0002 0.0002 -

DİKİLİ -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -0.7359

AKHİSAR -0.0002 - - - - -

KUŞADASI -0.0002 -0.0003 - - - -

DİDİM -0.0003 -0.0004 -0.0003 -0.0003 -0.0003 -0.6720

BODRUM -0.0003 -0.0003 -0.0003 -0.0003 -0.0003 -0.8070

DALAMAN -0.0004 -0.0005 -0.0005 -0.0006 -0.0006 -2.9368

ANAMUR -0.0003 -0.0004 -0.0006 -0.0007 -0.0007 -1.9160

SİLİFKE -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -0.8126

İSKENDERUN - - - - - -

FİNİKE -0.0004 -0.0006 -0.0009 -0.0010 -0.0010 -6.2084

KAŞ -0.0003 -0.0006 -0.0008 -0.0010 -0.0010 -2.7616

SALİHLİ -0.0001 - - 0.0002 0.0002 -

SEFERİHİSAR -0.0002 - - - - -

ÖDEMİŞ -0.0003 -0.0003 -0.0002 -0.0002 -0.0002 -

NAZİLLİ - - - - - -

ELMALI - 0.0003 0.0005 0.0007 0.0007 0.2513

MUT -0.0001 - - - - -

KARATAŞ -0.0003 -0.0005 -0.0006 -0.0007 -0.0007 -1.2930

MENEMEN -0.0003 -0.0003 -0.0002 - - -

FETHİYE -0.0003 -0.0005 -0.0006 -0.0007 -0.0007 -1.7236

MARMARİS -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -0.9714

BURHANİYE -0.0002 - - - - -

MİLAS -0.0002 -0.0003 -0.0002 -0.0002 -0.0002 -

YATAĞAN -0.0002 - - - - -

KOZAN -0.0003 -0.0005 -0.0005 -0.0006 -0.0006 -1.6484

132

Table 11. (cont’d) Trend slope values for model 1-1


DATÇA -0.0004 -0.0006 -0.0008 -0.0009 -0.0009 -3.2641

KÖYCEĞİZ -0.0003 -0.0004 -0.0004 -0.0004 -0.0004 -1.5417

KORKUTELİ -0.0003 -0.0005 -0.0006 -0.0007 -0.0007 -1.4587

KARAİSALI -0.0003 -0.0005 -0.0007 -0.0008 -0.0008 -1.6726

MANAVGAT -0.0002 -0.0004 -0.0005 -0.0006 -0.0006 -2.3695

ERDEMLİ -0.0002 -0.0004 -0.0006 -0.0007 -0.0007 -1.0295

CEYHAN -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.0131

DÖRTYOL - - -0.0003 -0.0004 -0.0004 -1.2068

ISLAHİYE -0.0002 -0.0002 -0.0003 -0.0003 -0.0003 -1.0821

GAZİPAŞA -0.0002 -0.0004 -0.0005 -0.0005 -0.0005 -1.9558

YUMURTALIK -0.0003 -0.0004 -0.0003 -0.0004 -0.0004 -1.1646

SAMANDAĞ -0.0002 -0.0004 -0.0005 -0.0006 -0.0006 -1.7947

ACIPAYAM - 0.0002 0.0004 0.0005 0.0005 0.2864

TEFENNİ -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -0.2395

GEMLİK - - - - - -

KARACABEY - - - - - -

MUDANYA - 0.0001 0.0003 0.0004 0.0004 0.6982

M.KEMALPAŞA - - - - - -

AYVACIK -0.0002 -0.0004 -0.0004 -0.0005 -0.0005 -1.2531

OSMANIYE -0.0001 -0.0002 -0.0003 -0.0004 -0.0004 -0.5273

ALANYA -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -1.9989

MANISA -0.0002 - - - - -

IZMIR -0.0002 -0.0002 -0.0002 - - -

AYDIN -0.0002 -0.0002 - - - -

DENIZLI - - 0.0003 0.0005 0.0005 0.5198

MUGLA -0.0004 -0.0006 -0.0007 -0.0008 -0.0008 -4.6311

ANTALYA -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.9478

MERSIN -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.0536

ADANA -0.0003 -0.0005 -0.0006 -0.0007 -0.0007 -1.1703

ANTAKYA -0.0003 -0.0004 -0.0004 -0.0005 -0.0005 -0.6076

BALIKESİR -0.0001 - - - - -

CANAKKALE -0.0001 - - - - -

BURSA -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -1.6586

133



BANDIRMA - - - 0.0000 - -

AYVALIK -0.0002 - - -0.0004 - -0.7978

DİKİLİ -0.0002 -0.0003 -0.0003 -0.0003 - -

AKHİSAR -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -0.8166

KUŞADASI -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.1836

DİDİM -0.0002 -0.0005 -0.0006 -0.0007 -0.0007 -1.2741

BODRUM -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.2539

DALAMAN -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -2.1662

ANAMUR -0.0002 -0.0004 -0.0006 -0.0007 -0.0007 -2.0738

SİLİFKE -0.0003 -0.0005 -0.0006 -0.0008 -0.0009 -1.8544

İSKENDERUN -0.0003 -0.0004 -0.0004 -0.0005 -0.0006 -0.8768

FİNİKE -0.0003 -0.0005 -0.0007 -0.0009 -0.0009 -3.2208

KAŞ -0.0003 -0.0005 -0.0007 -0.0008 -0.0009 -2.2175

SALİHLİ -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.1015

SEFERİHİSAR -0.0002 -0.0004 -0.0005 -0.0005 -0.0006 -0.7468

ÖDEMİŞ -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.3725

NAZİLLİ -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.0521

ELMALI -0.0005 -0.0007 -0.0010 -0.0011 -0.0012 -2.1464

MUT -0.0003 -0.0006 -0.0007 -0.0008 -0.0009 -1.1675

KARATAŞ -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -0.8206

MENEMEN -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -0.9390

FETHİYE -0.0003 -0.0005 -0.0005 -0.0006 -0.0007 -2.0024

MARMARİS -0.0003 -0.0005 -0.0007 -0.0008 -0.0008 -2.4706

BURHANİYE -0.0002 - - -0.0004 - -

MİLAS -0.0003 -0.0005 -0.0007 -0.0008 -0.0009 -1.6289

YATAĞAN -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -1.7983

KOZAN -0.0001 -0.0002 - -0.0003 - -

DATÇA -0.0003 -0.0005 -0.0007 -0.0008 -0.0008 -1.2993

KÖYCEĞİZ -0.0003 -0.0006 -0.0008 -0.0009 -0.0010 -2.6385

KORKUTELİ -0.0005 -0.0008 -0.0010 -0.0012 -0.0013 -2.6246

KARAİSALI -0.0002 -0.0003 -0.0005 -0.0005 -0.0006 -1.7505

MANAVGAT -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -1.5245

ERDEMLİ -0.0002 -0.0004 -0.0005 -0.0006 -0.0007 -1.5895

134



CEYHAN -0.0001 -0.0002 -0.0003 -0.0003 -0.0003 -

DÖRTYOL -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.2645

ISLAHİYE -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.0859

GAZİPAŞA - -0.0003 -0.0004 -0.0004 -0.0004 -1.1386

YUMURTALIK -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -

SAMANDAĞ -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -

ACIPAYAM -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -1.2656

TEFENNİ -0.0004 -0.0006 -0.0008 -0.0010 -0.0011 -1.5683

GEMLİK - - - 0.0001 - -

KARACABEY - - - -0.0001 - -

MUDANYA - - - 0.0002 0.0003 -

M.KEMALPAŞA - - - -0.0001 - -

AYVACIK -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -0.8111

OSMANIYE -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.9943

ALANYA -0.0002 -0.0004 -0.0005 -0.0005 -0.0006 -1.7941

MANISA -0.0003 -0.0005 -0.0006 -0.0007 -0.0007 -1.0241

IZMIR -0.0003 -0.0005 -0.0006 -0.0007 -0.0007 -1.2277

AYDIN -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.2547

DENIZLI -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.0201

MUGLA -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -2.5862

ANTALYA -0.0002 -0.0004 -0.0006 -0.0007 -0.0008 -3.9625

MERSIN -0.0002 -0.0004 -0.0005 -0.0006 -0.0007 -1.7131

ADANA -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.8752

ANTAKYA -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -0.9184

BALIKESİR -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -0.6535

CANAKKALE -0.0002 - - -0.0002 - -

BURSA - - - 0.0000 - -

135



BANDIRMA - - - - - -

AYVALIK - - - - - -

DİKİLİ - - - - - -

AKHİSAR -0.0001 - - - - -

KUŞADASI -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -0.9031

DİDİM -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.9960

BODRUM -0.0002 -0.0004 -0.0005 -0.0006 -0.0007 -1.4405

DALAMAN -0.0002 -0.0004 -0.0005 -0.0006 -0.0006 -1.8145

ANAMUR - -0.0002 -0.0003 -0.0003 -0.0004 -

SİLİFKE -0.0002 -0.0003 -0.0004 -0.0004 -0.0004 -0.8435


FİNİKE -0.0003 -0.0005 -0.0007 -0.0008 -0.0009 -3.2487

KAŞ -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -2.2196

SALİHLİ -0.0002 -0.0002 - - - -

SEFERİHİSAR -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -0.4410

ÖDEMİŞ -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -0.8051

NAZİLLİ -0.0001 -0.0002 -0.0002 -0.0002 - -

ELMALI -0.0003 -0.0005 -0.0005 -0.0006 -0.0007 -0.8198

MUT -0.0001 -0.0002 - - - -

KARATAŞ -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -0.7611

MENEMEN -0.0001 -0.0002 - - - -

FETHİYE -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -1.1707

MARMARİS -0.0003 -0.0005 -0.0007 -0.0008 -0.0008 -2.7955

BURHANİYE - - - - - -

MİLAS -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.0603

YATAĞAN -0.0002 -0.0003 -0.0004 -0.0004 -0.0004 -0.7799

KOZAN - - - - -0.0002 -

DATÇA -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.5281

KÖYCEĞİZ -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -2.2773

KORKUTELİ -0.0004 -0.0005 -0.0007 -0.0007 -0.0008 -1.8245

KARAİSALI - -0.0002 -0.0002 -0.0003 -0.0003 -0.6273

MANAVGAT - -0.0002 -0.0003 -0.0003 -0.0003 -1.0409

ERDEMLİ - - - -0.0001 - -

136



CEYHAN -0.0002 -0.0002 -0.0002 -0.0002 -0.0003 -

DÖRTYOL - - - - - -

ISLAHİYE -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -0.7705

GAZİPAŞA - -0.0002 -0.0002 - - -

YUMURTALIK - - - - - -

SAMANDAĞ -0.0001 -0.0002 -0.0002 -0.0002 - -

ACIPAYAM -0.0002 -0.0003 -0.0004 -0.0004 -0.0004 -0.3892

TEFENNİ -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.3714

GEMLİK - - 0.0002 0.0003 0.0003 -


MUDANYA - - 0.0003 0.0004 0.0004 0.6907


AYVACIK -0.0001 -0.0002 - - -0.0002 -

OSMANIYE - - - - - -

ALANYA - - -0.0002 -0.0003 -0.0003 -

MANISA -0.0001 - - - - -

IZMIR -0.0002 -0.0002 -0.0002 - - -

AYDIN -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -

DENIZLI -0.0001 -0.0002 - - - -

MUGLA -0.0003 -0.0005 -0.0006 -0.0006 -0.0007 -2.6603

ANTALYA -0.0002 -0.0003 -0.0005 -0.0005 -0.0006 -2.5428

MERSIN - -0.0001 -0.0002 -0.0003 -0.0003 -0.6400

ADANA -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -0.7370

ANTAKYA - -0.0002 -0.0002 - - -

BALIKESİR - - - - - -

CANAKKALE - - - - - -

BURSA - - - - - -

137




AYVALIK -0.0002 -0.0003 -0.0003 -0.0003 -0.0003 -

DİKİLİ -0.0002 -0.0002 -0.0002 - - -

AKHİSAR - -0.0003 -0.0003 -0.0003 -0.0003 -

KUŞADASI -0.0003 -0.0005 -0.0005 -0.0006 -0.0006 -1.0224

DİDİM -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.4823

BODRUM -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.3338

DALAMAN -0.0003 -0.0006 -0.0007 -0.0008 -0.0009 -2.8391

ANAMUR -0.0001 -0.0003 -0.0004 -0.0005 -0.0006 -0.9105

SİLİFKE - -0.0002 -0.0003 -0.0004 -0.0004 -0.6056

İSKENDERUN -0.0001 - - - - -

FİNİKE -0.0004 -0.0007 -0.0009 -0.0011 -0.0012 -3.6544

KAŞ -0.0003 -0.0006 -0.0008 -0.0009 -0.0010 -2.2666

SALİHLİ -0.0003 -0.0005 -0.0007 -0.0008 -0.0009 -0.9584

SEFERİHİSAR - -0.0004 -0.0005 -0.0005 -0.0006 -0.4760

ÖDEMİŞ -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -0.8065

NAZİLLİ - -0.0005 -0.0006 -0.0006 -0.0007 -0.4756

ELMALI - -0.0004 -0.0005 -0.0005 -0.0006 -0.3239

MUT - -0.0002 -0.0002 - - -

KARATAŞ -0.0001 -0.0002 -0.0003 -0.0003 -0.0004 -0.7029

MENEMEN -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -1.0610

FETHİYE -0.0003 -0.0005 -0.0007 -0.0007 -0.0008 -1.8234

MARMARİS -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -2.6772

BURHANİYE - -0.0003 -0.0003 -0.0003 -0.0003 -

MİLAS - -0.0004 -0.0005 -0.0005 -0.0006 -0.8210

YATAĞAN - -0.0006 -0.0007 -0.0008 -0.0009 -0.9778

KOZAN -0.0001 -0.0002 - - - -

DATÇA -0.0004 -0.0007 -0.0009 -0.0011 -0.0012 -3.9260

KÖYCEĞİZ - -0.0005 -0.0007 -0.0007 -0.0008 -1.9815

KORKUTELİ -0.0003 -0.0005 -0.0006 -0.0008 -0.0009 -2.1072

KARAİSALI -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -0.9432

MANAVGAT -0.0002 -0.0003 -0.0004 -0.0004 -0.0004 -1.4669

ERDEMLİ -0.0001 -0.0002 -0.0003 -0.0004 -0.0005 -0.5452

138



CEYHAN -0.0002 - -0.0003 -0.0003 - -

DÖRTYOL - - - - - -

ISLAHİYE -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.7000

GAZİPAŞA -0.0001 -0.0002 - - - -

YUMURTALIK -0.0001 - - - - -

SAMANDAĞ -0.0001 - -0.0003 - - -

ACIPAYAM -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -0.4870

TEFENNİ -0.0003 -0.0005 -0.0006 -0.0006 -0.0007 -0.7070

GEMLİK -0.0001 - - - - -


MUDANYA - - - - - -

M.KEMALPAŞA -0.0001 -0.0001 - - - -

AYVACIK -0.0003 -0.0005 -0.0005 -0.0006 -0.0007 -1.6756


ALANYA -0.0001 -0.0003 -0.0003 -0.0003 -0.0004 -

MANISA - -0.0004 -0.0006 -0.0007 -0.0007 -0.6581

IZMIR -0.0003 -0.0004 -0.0005 -0.0005 -0.0005 -0.8163

AYDIN - -0.0004 -0.0005 -0.0005 -0.0006 -0.7465

DENIZLI -0.0003 -0.0005 -0.0006 -0.0006 -0.0007 -0.5057

MUGLA -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -3.3837

ANTALYA -0.0003 -0.0006 -0.0007 -0.0008 -0.0009 -2.3454

MERSIN - -0.0002 -0.0003 -0.0004 -0.0004 -0.5228

ADANA -0.0002 -0.0002 -0.0003 - - -0.6302

ANTAKYA -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -0.3788

BALIKESİR -0.0002 -0.0002 - - - -

CANAKKALE -0.0002 -0.0003 - - - -

BURSA -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -1.1386

139




AYVALIK - - - - - -

DİKİLİ - - - - - -

AKHİSAR -0.0002 -0.0002 - - - -

KUŞADASI -0.0002 - - - - -

DİDİM -0.0001 - - - - -

BODRUM -0.0001 - - - - -0.4787

DALAMAN -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -0.7803

ANAMUR -0.0001 -0.0002 -0.0004 -0.0005 -0.0005 -1.1723

SİLİFKE - - -0.0002 -0.0002 -0.0003 -


FİNİKE - - - - - -

KAŞ -0.0002 -0.0004 -0.0006 -0.0008 -0.0009 -1.8069

SALİHLİ -0.0001 - - - - -

SEFERİHİSAR - - - - - -

ÖDEMİŞ -0.0003 -0.0004 -0.0005 -0.0007 -0.0007 -1.4170

NAZİLLİ -0.0001 - - - - -

ELMALI -0.0002 -0.0003 -0.0005 -0.0006 -0.0007 -0.8330

MUT - - - - - -

KARATAŞ - - - - - -

MENEMEN -0.0001 - - - -0.0003 -

FETHİYE -0.0002 -0.0004 -0.0005 -0.0007 -0.0007 -1.0119

MARMARİS -0.0002 -0.0002 -0.0003 -0.0004 -0.0005 -1.3780


MİLAS -0.0002 -0.0002 -0.0003 -0.0004 -0.0004 -0.5783

YATAĞAN -0.0001 - - - - -

KOZAN - - - - - -

DATÇA -0.0001 - - - - -

KÖYCEĞİZ -0.0003 -0.0004 -0.0006 -0.0007 -0.0008 -1.2032

KORKUTELİ - - - - - -

KARAİSALI - - - -0.0003 -0.0003 -

MANAVGAT - - - - - -0.6980

ERDEMLİ - - - -0.0003 -0.0004 -

140



CEYHAN - - - - - -

DÖRTYOL -0.0001 -0.0002 -0.0003 -0.0003 -0.0004 -0.3883

ISLAHİYE -0.0001 -0.0002 - -0.0003 -0.0003 -

GAZİPAŞA - - - - - -

YUMURTALIK - - 0.0002 0.0002 0.0003 -

SAMANDAĞ - 0.0002 0.0003 0.0004 0.0005 1.9858

ACIPAYAM -0.0002 -0.0002 - - - -

TEFENNİ -0.0002 - - - - -0.4306

GEMLİK - - - - - -


MUDANYA 0.0002 0.0004 0.0007 0.0008 0.0009 2.4122


AYVACIK - - - - - -0.5220

OSMANIYE -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -0.5156

ALANYA - - - - - -

MANISA - - - - - -

IZMIR - - - - - -

AYDIN -0.0002 -0.0002 -0.0003 -0.0003 -0.0003 -0.4104

DENIZLI -0.0001 - - - - -

MUGLA -0.0003 -0.0005 -0.0006 -0.0008 -0.0009 -2.6566

ANTALYA -0.0002 -0.0004 -0.0005 -0.0007 -0.0007 -1.5869

MERSIN 0.0001 0.0002 0.0003 0.0004 0.0004 0.6073

ADANA - - - - - -

ANTAKYA - - - - - -



BURSA - - - - - -

141



BANDIRMA - - 0.0002 0.0003 0.0003 0.4893

AYVALIK - - 0.0002 0.0002 0.0002 -

DİKİLİ - - - - 0.0002 -

AKHİSAR - - - - - -

KUŞADASI -0.0001 - - - - -

DİDİM - - - - - -

BODRUM -0.0001 - -0.0001 - - -

DALAMAN -0.0001 -0.0001 -0.0002 - - -

ANAMUR -0.0001 -0.0002 -0.0002 -0.0002 -0.0003 -

SİLİFKE -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -1.4521

İSKENDERUN -0.0002 -0.0002 -0.0003 -0.0004 -0.0004 -0.7100

FİNİKE -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -2.0834

KAŞ -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.4103

SALİHLİ -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -

SEFERİHİSAR -0.0001 - - - - -

ÖDEMİŞ -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.1444

NAZİLLİ -0.0002 -0.0002 -0.0003 -0.0004 -0.0004 -0.9157

ELMALI -0.0004 -0.0006 -0.0007 -0.0009 -0.0010 -1.8405

MUT -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.1899

KARATAŞ - - - - - -

MENEMEN -0.0001 - - - - -

FETHİYE -0.0001 -0.0002 -0.0002 -0.0002 -0.0003 -1.1177

MARMARİS -0.0001 -0.0002 -0.0002 - - -


MİLAS -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -1.1883

YATAĞAN -0.0003 -0.0004 -0.0006 -0.0007 -0.0008 -1.7752

KOZAN - - - - - -

DATÇA -0.0001 -0.0001 - - - -

KÖYCEĞİZ -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -2.2099

KORKUTELİ -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -2.1082

KARAİSALI - - - - - -

MANAVGAT - - - - - -

ERDEMLİ -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -

142



CEYHAN - - - - - -

DÖRTYOL -0.0002 -0.0002 -0.0002 -0.0003 -0.0003 -0.7822

ISLAHİYE -0.0003 -0.0004 -0.0004 -0.0005 -0.0006 -1.0588


YUMURTALIK - - - - - -

SAMANDAĞ -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -0.9092

ACIPAYAM -0.0004 -0.0005 -0.0007 -0.0008 -0.0009 -1.4951

TEFENNİ -0.0004 -0.0005 -0.0006 -0.0008 -0.0009 -1.3775

GEMLİK - 0.0002 0.0003 0.0004 0.0004 0.8583

KARACABEY - - 0.0003 0.0004 0.0004 0.7794

MUDANYA 0.0002 0.0003 0.0005 0.0006 0.0007 1.5519

M.KEMALPAŞA - - 0.0003 0.0003 0.0004 0.7508

AYVACIK -0.0001 -0.0001 -0.0002 -0.0002 -0.0002 -

OSMANIYE - -0.0001 -0.0002 -0.0002 - -

ALANYA -0.0001 - - - - -

MANISA -0.0001 - - - - -

IZMIR -0.0001 -0.0002 - - - -

AYDIN -0.0002 -0.0002 -0.0002 -0.0002 -0.0003 -

DENIZLI -0.0003 -0.0004 -0.0006 -0.0007 -0.0008 -1.4013

MUGLA -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -2.3274

ANTALYA -0.0002 -0.0002 -0.0003 -0.0004 -0.0004 -2.9119

MERSIN - - - - - -

ADANA - - - - - -

ANTAKYA -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.7609



BURSA - - - - - -

143



BANDIRMA 0.0001 0.0002 0.0004 0.0005 0.0006 0.8874

AYVALIK - - 0.0001 0.0002 0.0002 -

DİKİLİ - - - - - -

AKHİSAR - - - - - -

KUŞADASI - - - - - -

DİDİM - - - - - -

BODRUM -0.0001 -0.0001 - - - -

DALAMAN -0.0002 -0.0002 -0.0003 -0.0003 -0.0003 -1.4235

ANAMUR - -0.0001 - - -0.0003 -

SİLİFKE -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.0499

İSKENDERUN -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -0.7376

FİNİKE -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.4041

KAŞ -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.0963

SALİHLİ -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -0.4387

SEFERİHİSAR -0.0001 - - - - -

ÖDEMİŞ - - - - - -

NAZİLLİ - - - - - -

ELMALI - - - - - -

MUT - -0.0002 -0.0002 -0.0003 -0.0004 -0.3133

KARATAŞ -0.0001 - - - - -

MENEMEN -0.0001 - - - - -

FETHİYE -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.2153

MARMARİS -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -1.4218

BURHANİYE - -0.0001 - -0.0002 - -

MİLAS -0.0001 -0.0002 -0.0002 -0.0003 - -0.6283

YATAĞAN -0.0002 -0.0003 -0.0005 -0.0006 -0.0006 -0.9051

KOZAN -0.0001 - - - - -

DATÇA -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -1.3057

KÖYCEĞİZ -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.7555

KORKUTELİ -0.0002 -0.0002 -0.0003 -0.0004 -0.0004 -

KARAİSALI -0.0001 -0.0001 -0.0002 -0.0003 -0.0003 -

MANAVGAT -0.0001 - - -0.0003 -0.0003 -1.4126

ERDEMLİ -0.0001 -0.0002 -0.0003 -0.0004 -0.0005 -0.7625

144



CEYHAN -0.0001 -0.0001 - - - -

DÖRTYOL - - - -0.0002 -0.0003 -

ISLAHİYE -0.0003 -0.0005 -0.0005 -0.0006 -0.0007 -1.8122


YUMURTALIK -0.0001 - - - - -

SAMANDAĞ -0.0002 -0.0002 -0.0003 -0.0003 -0.0003 -0.8575

ACIPAYAM -0.0001 - - - - -

TEFENNİ -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.5122

GEMLİK - - - - - -

KARACABEY - 0.0002 0.0004 0.0005 0.0006 0.6831

MUDANYA - - - 0.0004 0.0004 -

M.KEMALPAŞA - 0.0001 0.0002 0.0003 0.0004 -

AYVACIK - - - - - -

OSMANIYE -0.0001 - - - - -

ALANYA -0.0001 - -0.0002 -0.0003 -0.0004 -1.4152

MANISA -0.0001 -0.0002 -0.0002 -0.0003 -0.0003 -0.3649

IZMIR -0.0001 - - -0.0002 -0.0003 -0.5513

AYDIN - - - - - -

DENIZLI -0.0001 - - - - -0.2599

MUGLA -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -3.0774

ANTALYA -0.0002 -0.0002 -0.0002 -0.0003 -0.0004 -0.9998

MERSIN -0.0001 - -0.0002 -0.0003 -0.0004 -

ADANA -0.0002 -0.0001 -0.0001 - - -

ANTAKYA -0.0002 -0.0002 -0.0004 -0.0005 -0.0005 -0.4665


CANAKKALE - 0.0002 0.0003 0.0003 0.0004 0.5070

BURSA - - - - - -

145



BANDIRMA - - - - - 0.0944

AYVALIK -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.1208

DİKİLİ -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -1.2501

AKHİSAR -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.9951

KUŞADASI -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -

DİDİM - -0.0002 - - - -

BODRUM - -0.0002 -0.0002 - - -0.3317

DALAMAN -0.0002 -0.0003 -0.0004 -0.0006 -0.0007 -2.5499

ANAMUR -0.0003 -0.0004 -0.0005 -0.0007 -0.0007 -3.2585

SİLİFKE -0.0004 -0.0005 -0.0005 -0.0007 -0.0008 -1.9831

İSKENDERUN -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.6765

FİNİKE -0.0004 -0.0004 -0.0005 -0.0007 -0.0008 -3.4102

KAŞ -0.0003 -0.0004 -0.0005 -0.0007 -0.0008 -2.5599

SALİHLİ -0.0003 -0.0004 -0.0004 -0.0005 -0.0006 -0.8362

SEFERİHİSAR - -0.0003 -0.0003 -0.0004 -0.0005 -0.7297

ÖDEMİŞ -0.0003 -0.0005 -0.0005 -0.0006 -0.0007 -1.3885

NAZİLLİ -0.0003 -0.0004 -0.0005 -0.0007 -0.0008 -1.5202

ELMALI -0.0005 -0.0006 -0.0008 -0.0010 -0.0011 -1.9958

MUT -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -1.3588

KARATAŞ -0.0002 -0.0003 -0.0003 -0.0004 -0.0005 -1.3051

MENEMEN -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.0065

FETHİYE -0.0002 -0.0003 -0.0004 -0.0006 -0.0007 -3.1756

MARMARİS -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -1.4533

BURHANİYE -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.7963

MİLAS -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -1.2424

YATAĞAN -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -1.7317

KOZAN -0.0003 -0.0003 -0.0004 -0.0005 -0.0006 -

DATÇA -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -0.6776

KÖYCEĞİZ -0.0003 -0.0005 -0.0006 -0.0008 -0.0009 -2.7808

KORKUTELİ -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -1.7738

KARAİSALI -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -2.7659

MANAVGAT -0.0003 -0.0003 -0.0004 -0.0005 -0.0006 -2.7792

ERDEMLİ -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -2.0348

146



CEYHAN -0.0002 -0.0003 -0.0003 -0.0004 -0.0005 -1.1572

DÖRTYOL -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -2.3914

ISLAHİYE -0.0005 -0.0006 -0.0007 -0.0008 -0.0009 -2.2514

GAZİPAŞA -0.0003 -0.0003 -0.0005 -0.0006 -0.0006 -2.6127

YUMURTALIK -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -1.6012

SAMANDAĞ -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.8287

ACIPAYAM -0.0004 -0.0006 -0.0007 -0.0009 -0.0010 -1.6232

TEFENNİ -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -

GEMLİK - - - - 0.0002 -

KARACABEY - - - - - 0.2343

MUDANYA - - 0.0003 0.0004 0.0004 -

M.KEMALPAŞA - - - - - 0.2164

AYVACIK -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.2506

OSMANIYE -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.9880

ALANYA -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -

MANISA -0.0002 -0.0002 -0.0002 -0.0003 -0.0003 -0.4944

IZMIR -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.2204

AYDIN -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.3892

DENIZLI -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -0.8954

MUGLA -0.0004 -0.0005 -0.0007 -0.0008 -0.0009 -3.1253

ANTALYA -0.0004 -0.0004 -0.0005 -0.0007 -0.0008 -6.8051

MERSIN -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -2.0581

ADANA -0.0002 -0.0004 -0.0004 -0.0005 -0.0006 -1.5713

ANTAKYA -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -

BALIKESİR -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -0.6198

CANAKKALE -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -

BURSA -0.0001 - - - 0.0001 -

147



BANDIRMA - 0.0002 0.0003 0.0003 0.0004 0.9017

AYVALIK -0.0001 - - - - -

DİKİLİ -0.0001 - - - -0.0002 -

AKHİSAR -0.0002 - - - -0.0002 -

KUŞADASI -0.0002 -0.0002 -0.0002 -0.0003 - -

DİDİM -0.0002 -0.0002 - - - -

BODRUM -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -

DALAMAN -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.9095

ANAMUR - -0.0002 -0.0003 -0.0004 -0.0005 -1.2869

SİLİFKE -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -1.4583

İSKENDERUN -0.0001 - - - - -

FİNİKE -0.0002 -0.0004 -0.0006 -0.0008 -0.0009 -4.1823

KAŞ -0.0003 -0.0004 -0.0006 -0.0008 -0.0009 -2.6461

SALİHLİ -0.0002 -0.0002 - -0.0002 -0.0002 -


ÖDEMİŞ -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.1531

NAZİLLİ -0.0001 - - - - -

ELMALI -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.4453

MUT -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -

KARATAŞ -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -1.9938

MENEMEN -0.0002 - - - -0.0002 -

FETHİYE -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -1.6547

MARMARİS -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -2.1082

BURHANİYE -0.0001 - - - -0.0002 -

MİLAS -0.0002 -0.0003 -0.0003 -0.0004 -0.0005 -1.2091

YATAĞAN -0.0002 -0.0002 -0.0003 -0.0003 -0.0004 -0.8074

KOZAN -0.0002 - -0.0002 - -0.0002 -

DATÇA -0.0002 -0.0003 -0.0004 -0.0004 -0.0005 -0.9445

KÖYCEĞİZ -0.0002 -0.0003 -0.0004 -0.0005 -0.0006 -2.4524

KORKUTELİ -0.0003 -0.0005 -0.0007 -0.0008 -0.0010 -2.6903

KARAİSALI -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.4494

MANAVGAT -0.0001 -0.0002 -0.0003 -0.0004 -0.0005 -1.9299

ERDEMLİ - -0.0001 -0.0002 -0.0002 -0.0003 -0.6527

148



CEYHAN -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.2249

DÖRTYOL - - - - - -

ISLAHİYE -0.0003 -0.0003 -0.0003 -0.0003 -0.0004 -1.2682

GAZİPAŞA - -0.0001 -0.0002 -0.0002 -0.0003 -

YUMURTALIK -0.0002 -0.0002 -0.0002 -0.0002 -0.0003 -

SAMANDAĞ -0.0003 -0.0003 -0.0004 -0.0004 -0.0005 -1.8372

ACIPAYAM -0.0002 -0.0002 -0.0002 -0.0003 -0.0003 -

TEFENNİ -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -0.4529

GEMLİK - 0.0002 0.0003 0.0004 0.0005 0.9707

KARACABEY - 0.0002 0.0003 0.0004 0.0005 0.7738

MUDANYA - 0.0002 0.0003 0.0004 0.0005 0.8824


AYVACIK -0.0001 -0.0002 -0.0002 -0.0002 -0.0002 -


ALANYA - - -0.0002 -0.0003 -0.0003 -1.1261

MANISA -0.0001 - - - - -

IZMIR -0.0002 -0.0002 -0.0002 -0.0003 -0.0003 -

AYDIN -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -0.9821

DENIZLI -0.0001 - - - - -

MUGLA -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -2.9318

ANTALYA -0.0002 -0.0004 -0.0006 -0.0007 -0.0008 -4.6493

MERSIN -0.0001 -0.0003 -0.0003 -0.0004 -0.0005 -1.3385

ADANA -0.0002 -0.0003 -0.0004 -0.0005 -0.0005 -1.5819

ANTAKYA -0.0003 -0.0003 -0.0004 -0.0005 -0.0006 -1.4510

BALIKESİR - - 0.0001 0.0001 0.0002 -


BURSA - - - - 0.0001 -

149




AYVALIK - -0.0002 - - - -

DİKİLİ - - - - - -

AKHİSAR - - - - - -

KUŞADASI -0.0002 -0.0003 -0.0004 -0.0004 -0.0004 -0.9115

DİDİM - -0.0003 -0.0003 - - -

BODRUM - -0.0003 -0.0003 -0.0004 -0.0004 -

DALAMAN -0.0002 -0.0004 -0.0005 -0.0006 -0.0006 -2.6239

ANAMUR -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.5644

SİLİFKE -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.3257

İSKENDERUN -0.0003 -0.0003 -0.0003 -0.0004 -0.0005 -1.1992

FİNİKE -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -3.0494

KAŞ -0.0003 -0.0004 -0.0006 -0.0007 -0.0008 -2.0672

SALİHLİ - -0.0003 -0.0003 -0.0004 -0.0004 -0.5154


ÖDEMİŞ - -0.0002 -0.0002 -0.0002 -0.0002 -

NAZİLLİ - - - - - -

ELMALI - -0.0003 -0.0004 -0.0004 -0.0005 -0.3509

MUT -0.0001 -0.0002 -0.0003 -0.0003 -0.0004 -0.2685

KARATAŞ -0.0003 -0.0004 -0.0004 -0.0005 -0.0006 -1.9121

MENEMEN - -0.0003 -0.0004 -0.0004 -0.0004 -

FETHİYE -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.9081

MARMARİS -0.0002 -0.0004 -0.0005 -0.0005 -0.0006 -1.7822

BURHANİYE - -0.0002 -0.0002 -0.0002 -0.0003 -

MİLAS - -0.0002 -0.0002 -0.0002 -0.0002 -0.4515

YATAĞAN - -0.0003 -0.0004 -0.0005 -0.0006 -0.7760

KOZAN -0.0003 -0.0005 -0.0006 -0.0006 -0.0007 -2.1659

DATÇA -0.0003 -0.0004 -0.0006 -0.0006 -0.0007 -2.6884

KÖYCEĞİZ - -0.0004 -0.0004 -0.0005 -0.0006 -1.6422

KORKUTELİ -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -2.0579

KARAİSALI -0.0004 -0.0006 -0.0007 -0.0008 -0.0008 -2.1444

MANAVGAT -0.0002 -0.0004 -0.0004 -0.0005 -0.0006 -2.2814

ERDEMLİ -0.0002 -0.0004 -0.0005 -0.0006 -0.0007 -1.1042

150



CEYHAN -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -1.4530

DÖRTYOL -0.0003 -0.0003 -0.0003 -0.0003 -0.0003 -

ISLAHİYE -0.0004 -0.0005 -0.0006 -0.0006 -0.0007 -2.3811

GAZİPAŞA -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -1.4562

YUMURTALIK -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -1.6826

SAMANDAĞ -0.0004 -0.0005 -0.0005 -0.0006 -0.0007 -1.7906

ACIPAYAM - - -0.0002 -0.0002 -0.0002 -

TEFENNİ -0.0002 -0.0003 -0.0003 -0.0004 -0.0004 -0.4668

GEMLİK -0.0001 - - - - -

KARACABEY - - - 0.0003 0.0003 -

MUDANYA - - 0.0001 - - -

M.KEMALPAŞA -0.0001 - - - - -

AYVACIK -0.0003 -0.0004 -0.0005 -0.0006 -0.0006 -1.7674

OSMANIYE -0.0003 -0.0003 -0.0002 - - -

ALANYA -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -2.1112

MANISA - -0.0002 -0.0002 -0.0003 -0.0003 -

IZMIR -0.0002 -0.0003 -0.0003 -0.0003 -0.0004 -0.6595

AYDIN - - - - - -

DENIZLI - -0.0003 -0.0003 -0.0003 -0.0003 -0.2515

MUGLA -0.0003 -0.0004 -0.0006 -0.0006 -0.0007 -3.1938

ANTALYA -0.0003 -0.0005 -0.0006 -0.0007 -0.0008 -1.9951

MERSIN -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -1.2421

ADANA -0.0003 -0.0004 -0.0005 -0.0005 -0.0006 -1.4225

ANTAKYA -0.0003 -0.0003 -0.0004 -0.0005 -0.0005 -0.4881

BALIKESİR -0.0002 - - - - -


BURSA -0.0001 -0.0001 - - - -

151



BANDIRMA -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -1.2155

AYVALIK -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -1.4128

DİKİLİ -0.0004 -0.0006 -0.0007 -0.0007 -0.0008 -1.5068

AKHİSAR -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -1.6499

KUŞADASI -0.0005 -0.0007 -0.0008 -0.0009 -0.0010 -1.5502

DİDİM -0.0005 -0.0008 -0.0009 -0.0010 -0.0011 -1.8023

BODRUM -0.0006 -0.0009 -0.0010 -0.0012 -0.0013 -1.8496

DALAMAN -0.0005 -0.0009 -0.0011 -0.0012 -0.0013 -3.9283

ANAMUR -0.0004 -0.0007 -0.0010 -0.0012 -0.0014 -2.0766

SİLİFKE -0.0004 -0.0007 -0.0010 -0.0012 -0.0013 -1.9075

İSKENDERUN -0.0005 -0.0008 -0.0010 -0.0012 -0.0013 -2.2981

FİNİKE -0.0006 -0.0009 -0.0011 -0.0013 -0.0014 -2.8071

KAŞ -0.0005 -0.0008 -0.0011 -0.0012 -0.0013 -2.1842

SALİHLİ -0.0005 -0.0007 -0.0009 -0.0011 -0.0012 -1.6147

SEFERİHİSAR -0.0004 -0.0006 -0.0008 -0.0009 -0.0009 -0.7805

ÖDEMİŞ -0.0005 -0.0007 -0.0009 -0.0011 -0.0013 -2.0717

NAZİLLİ -0.0003 -0.0005 -0.0008 -0.0009 -0.0010 -0.6492

ELMALI -0.0004 -0.0007 -0.0009 -0.0010 -0.0011 -0.5104

MUT -0.0003 -0.0007 -0.0009 -0.0011 -0.0012 -0.6598

KARATAŞ -0.0005 -0.0008 -0.0010 -0.0012 -0.0013 -1.8939

MENEMEN -0.0005 -0.0007 -0.0009 -0.0010 -0.0011 -1.9823

FETHİYE -0.0005 -0.0008 -0.0010 -0.0011 -0.0013 -2.4982

MARMARİS -0.0006 -0.0009 -0.0012 -0.0013 -0.0014 -3.1877

BURHANİYE -0.0004 -0.0005 -0.0006 -0.0007 -0.0007 -0.7858

MİLAS -0.0004 -0.0007 -0.0008 -0.0009 -0.0010 -1.5322

YATAĞAN -0.0005 -0.0007 -0.0010 -0.0011 -0.0013 -1.5941

KOZAN -0.0006 -0.0008 -0.0011 -0.0012 -0.0013 -2.9038

DATÇA -0.0006 -0.0010 -0.0012 -0.0013 -0.0014 -3.5946

KÖYCEĞİZ -0.0005 -0.0007 -0.0010 -0.0011 -0.0012 -3.0215

KORKUTELİ -0.0005 -0.0007 -0.0009 -0.0011 -0.0011 -2.3294

KARAİSALI -0.0005 -0.0009 -0.0011 -0.0012 -0.0013 -2.0062

MANAVGAT -0.0004 -0.0007 -0.0010 -0.0012 -0.0013 -3.4744

ERDEMLİ -0.0003 -0.0005 -0.0009 -0.0011 -0.0013 -1.1609

152



CEYHAN -0.0006 -0.0009 -0.0010 -0.0012 -0.0013 -1.8455

DÖRTYOL -0.0004 -0.0006 -0.0008 -0.0010 -0.0011 -2.2443

ISLAHİYE -0.0006 -0.0010 -0.0012 -0.0014 -0.0016 -3.7407

GAZİPAŞA -0.0004 -0.0007 -0.0010 -0.0012 -0.0013 -3.0907

YUMURTALIK -0.0006 -0.0009 -0.0011 -0.0012 -0.0013 -2.2480

SAMANDAĞ -0.0006 -0.0009 -0.0011 -0.0013 -0.0014 -3.0354

ACIPAYAM -0.0004 -0.0007 -0.0008 -0.0010 -0.0011 -0.8360

TEFENNİ -0.0005 -0.0007 -0.0010 -0.0011 -0.0013 -1.2053

GEMLİK -0.0005 -0.0007 -0.0009 -0.0010 -0.0012 -2.8118

KARACABEY -0.0003 -0.0004 -0.0005 -0.0006 -0.0007 -0.6614

MUDANYA -0.0004 -0.0005 -0.0006 -0.0007 -0.0008 -1.4393

M.KEMALPAŞA -0.0004 -0.0006 -0.0007 -0.0008 -0.0009 -1.2135

AYVACIK -0.0005 -0.0008 -0.0010 -0.0011 -0.0012 -2.7502

OSMANIYE -0.0003 -0.0005 -0.0007 -0.0009 -0.0010 -1.3446

ALANYA -0.0004 -0.0007 -0.0010 -0.0011 -0.0012 -3.5498

MANISA -0.0004 -0.0007 -0.0009 -0.0010 -0.0011 -1.1182

IZMIR -0.0004 -0.0007 -0.0008 -0.0009 -0.0010 -1.7293

AYDIN -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -1.2404

DENIZLI -0.0004 -0.0007 -0.0009 -0.0011 -0.0012 -0.8129

MUGLA -0.0007 -0.0010 -0.0012 -0.0013 -0.0015 -5.6805

ANTALYA -0.0005 -0.0008 -0.0010 -0.0012 -0.0013 -1.8437

MERSIN -0.0003 -0.0006 -0.0009 -0.0011 -0.0012 -1.1522

ADANA -0.0006 -0.0009 -0.0011 -0.0012 -0.0014 -2.0974

ANTAKYA -0.0005 -0.0008 -0.0010 -0.0012 -0.0013 -0.8290

BALIKESİR -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -1.0592

CANAKKALE -0.0002 -0.0003 -0.0004 -0.0004 -0.0004 -

BURSA -0.0005 -0.0008 -0.0010 -0.0012 -0.0013 -4.2393

153



BANDIRMA 0.0003 0.0004 0.0006 0.0009 0.0010 2.1309

AYVALIK 0.0003 0.0006 0.0008 0.0011 0.0012 4.0884

DİKİLİ 0.0005 0.0008 0.0012 0.0015 0.0016 11.1233

AKHİSAR - - 0.0003 0.0004 0.0004 -

KUŞADASI 0.0003 0.0004 0.0007 0.0010 0.0012 4.3631

DİDİM 0.0001 0.0002 0.0004 0.0006 0.0006 2.5433

BODRUM - - - - - -

DALAMAN 0.0001 0.0002 0.0004 0.0006 0.0008 3.3405

ANAMUR -0.0002 -0.0003 -0.0005 -0.0007 -0.0008 -1.8209

SİLİFKE -0.0003 -0.0005 -0.0006 -0.0008 -0.0009 -1.1679

İSKENDERUN -0.0003 -0.0004 -0.0006 -0.0007 -0.0008 -1.2993

FİNİKE -0.0003 -0.0004 -0.0006 -0.0008 -0.0010 -2.7210

KAŞ -0.0002 -0.0002 -0.0004 -0.0005 -0.0007 -1.9564

SALİHLİ -0.0004 -0.0006 -0.0008 -0.0009 -0.0010 -2.2055

SEFERİHİSAR 0.0001 0.0002 0.0004 0.0005 0.0006 1.2161

ÖDEMİŞ - - - - - -

NAZİLLİ -0.0001 - -0.0003 -0.0003 -0.0004 -

ELMALI -0.0006 -0.0008 -0.0010 -0.0012 -0.0014 -2.7304

MUT -0.0004 -0.0006 -0.0008 -0.0010 -0.0011 -1.0511

KARATAŞ - - - - - -0.9660

MENEMEN 0.0003 0.0005 0.0008 0.0010 0.0012 5.6682

FETHİYE 0.0000 - 0.0002 0.0003 0.0004 1.5669

MARMARİS - - - - - -

BURHANİYE 0.0003 0.0006 0.0008 0.0011 0.0012 3.7722

MİLAS - 0.0002 0.0004 0.0006 0.0007 3.4625

YATAĞAN -0.0005 -0.0007 -0.0010 -0.0012 -0.0013 -2.8793

KOZAN -0.0001 - - - - -

DATÇA - - - - - -

KÖYCEĞİZ - - 0.0002 0.0003 0.0004 2.5084

KORKUTELİ -0.0006 -0.0008 -0.0010 -0.0012 -0.0014 -3.5397

KARAİSALI -0.0002 - -0.0003 -0.0004 -0.0004 -1.0600

MANAVGAT 0.0001 0.0002 0.0005 0.0007 0.0009 4.2760

ERDEMLİ -0.0003 -0.0004 -0.0005 -0.0007 -0.0009 -1.0033

154



CEYHAN -0.0002 - - - - -

DÖRTYOL 0.0002 0.0003 0.0004 0.0006 0.0007 3.3084

ISLAHİYE -0.0006 -0.0009 -0.0012 -0.0014 -0.0015 -5.0346

GAZİPAŞA - 0.0002 0.0003 0.0005 0.0006 1.8991

YUMURTALIK -0.0001 - - - - -

SAMANDAĞ -0.0002 -0.0004 -0.0005 -0.0007 -0.0008 -4.5645

ACIPAYAM -0.0006 -0.0008 -0.0010 -0.0012 -0.0013 -2.1494

TEFENNİ -0.0005 -0.0008 -0.0010 -0.0011 -0.0012 -1.9704

GEMLİK 0.0003 0.0005 0.0006 0.0008 0.0009 4.1890

KARACABEY 0.0001 0.0003 0.0004 0.0005 0.0005 1.0010

MUDANYA 0.0002 0.0004 0.0006 0.0007 0.0008 1.8873


AYVACIK 0.0003 0.0006 0.0009 0.0012 0.0014 6.0496


ALANYA 0.0001 0.0002 0.0003 0.0005 0.0006 2.0866

MANISA -0.0002 -0.0003 -0.0004 -0.0006 -0.0007 -0.9838

IZMIR 0.0002 0.0004 0.0006 0.0009 0.0010 5.2708

AYDIN - - - - - -

DENIZLI -0.0006 -0.0008 -0.0011 -0.0013 -0.0014 -2.6969

MUGLA - - - - - -

ANTALYA -0.0002 -0.0002 -0.0003 -0.0004 -0.0005 -1.3877

MERSIN -0.0002 -0.0003 -0.0004 -0.0005 -0.0007 -0.8997

ADANA -0.0001 - - - - -

ANTAKYA -0.0004 -0.0007 -0.0010 -0.0012 -0.0014 -2.3972


CANAKKALE 0.0004 0.0007 0.0010 0.0013 0.0014 4.1352

BURSA - - - - - -

Documents

DROUGHT ANALYSIS USING CORDEX SIMULATIONS ...etd.lib.metu.edu.tr/upload/12622080/index.pdfTHE DEGREE OF MASTER OF SCIENCE IN CIVIL ENGINEERING APRIL 2018 iii Approval of the thesis: